fullseye

文献層: 運転のシミュレーションと評価

何のコーパスか: シミュレータ(CARLA・MetaDrive・AWSIM・SUMO 連成・デジタルツイン)、シナリオに基づく試験(OpenSCENARIO・危険度指標・コーナーケース・危険シナリオ生成)、稀な事象の加速評価(重要度サンプリング)、閉ループ評価(nuPlan・NAVSIM・Bench2Drive)、ニューラル再構成(NeRF・3DGS)、sim-to-real(ドメインギャップ・悪天候・合成データの真値)。

Fullseye でなぜ要るか: 運転 PoC の系列(opsdrive の 22 カテゴリ)を評価の作法につなぐ。どの量を真値で採点できるか、閉じた商用スタックが何を握っているか、自前の世界と CARLA を同じ場面で比べるときに何を測るか。

来歴: OpenAlex から取得した論文メタデータ 1,197 本(2012 年以降、fetch_openalex_topical.py)を TF-IDF + k-means で 62 クラスタに階層化し(raptor の corpus2skill、2026-10-03)、各クラスタの要約を LLM(claude-haiku-4-5)が書いた。ここに載せるのはその要約と、代表論文の題名・著者・年・DOI だけ(メタデータは CC0。抄録は出版社の権利なので写さない)。要約は生成時の英語のまま。正直な限界: 文献は原理・選定・校正の研究が中心で、型番やデータシートの数値(製品知識)は薄い。

使い方(op に落とす): 各クラスタの「使う op」は、知識のテーマ → op の対応表(tools/gen_literature_notes.py の THEME_OPS、人が書いたもの)で引いた実在の op(生成時に出荷ノート 3,623 本の一覧と突き合わせ、無ければ止まる)。op の後ろはなぜ使うかと型。テーマに当たらないクラスタは「op に落ちていない」と書く(Fullseye に無い領域を隠さない)。読む順: 工程・部品の話題 → クラスタ → 使う op → op ノート(型の契約・実行できる例)→ 実装。末尾の「op → クラスタ」で逆に引ける。

目次

• sampling / importance / rare(57 本)
• traffic / simulation / vehicle(114 本)
• scenario / testing / automated(210 本)
• weather / detection / synthetic(178 本)
• digital / twin / industry(57 本)
• planning / driving / learning(217 本)
• reconstruction / scene / view(132 本)
• driving / learning / autonomous(232 本)

sampling / importance / rare

57 本

Statistical machinery for estimating very small failure probabilities of autonomous systems in simulation: importance sampling, adaptive splitting, rare-event Monte Carlo and accelerated evaluation of automated vehicles. Methods range from piecewise-mixture and Kriging surrogates fitted to naturalistic driving data to deep, normalizing-flow and diffusion-based proposals, with some rare-transition theory borrowed from physics and climate science.

• Accelerated evaluation: Skew scenario parameters (cut-in gap, relative speed) toward risk, then reweight by likelihood ratios to recover unbiased naturalistic crash rates.
• Relative error / efficiency: Judge estimators by relative half-width and runs needed versus crude Monte Carlo; reported speedups often span 10^2-10^5.
• Learned proposals: Deep importance sampling, normalizing flows and diffusion models build proposal densities for high-dimensional black-box AV tests; some give certifiable bounds.
• Splitting and temporal monitors: Multilevel/adaptive splitting with reusable temporal-logic monitors estimates rare traffic-rule violations over long horizons.
• Surrogate models: Kriging and kernel methods cut expensive on-track or simulator runs by modelling the performance boundary.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• 稀な事象の加速評価(重要度サンプリング): importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)

rare / algorithm / events

A loose leaf linking rare events to learning and reliability algorithms: importance-weighted prioritized replay in SAC for obstacle avoidance, importance-sampled MCTS, RL for locating transition barriers, and fault-vulnerability estimation for DNN accelerators.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Yulin Ma et al. (2025). Intelligent obstacle avoidance control method for autonomous vehicles based on improved SAC algorithm. 10.1504/ijvd.2025.151524
  • You Zhou et al. (2025). Communication Efficient Cooperative Edge AI via Event-Triggered Computation Offloading. 10.48550/arxiv.2501.02001
  • Adittya Pal (2024). Estimating reaction barriers with deep reinforcement learning. 10.3233/ds-240063
  • Tianxiang Ou et al. (2024). ImS-MCTS: A New Importance Sampling MCTS Algorithm for Discrete Event System. 10.3233/atde240143
  • Tyagi et al. (2022). Thales: Formulating and Estimating Architectural Vulnerability Factors for DNN Accelerators. 10.48550/arxiv.2212.02649
  • Marie Siew et al. (2022). FIRE: A Failure-Adaptive Reinforcement Learning Framework for Edge Computing Migrations. 10.48550/arxiv.2209.14399

markov / carlo / monte

Rare-event estimation and safety verification built on Markov models and Monte Carlo: behavioural event-rate estimation for AV evaluation, end-to-end rare-event simulation of AV stacks, model-free safety verification of MDPs and sequential Monte Carlo data assimilation.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Yujia Chen et al. (2026). Betting for Sim-to-Real Performance Evaluation. 10.15607/rss.2026.xxii.090
  • Abhijit Mazumdar et al. (2024). Online Model-Free Safety Verification for Markov Decision Processes Without Safety Violation. 10.23919/ecc64448.2024.10590743
  • Abhijit Mazumdar et al. (2023). Online Model-free Safety Verification for Markov Decision Processes Without Safety Violation. 10.48550/arxiv.2312.05243
  • Maria A. Terres et al. (2023). Behavioral event detection and rate estimation for autonomous vehicle evaluation. 10.1002/asmb.2769
  • Haidong Xue (2022). Data Assimilation Based on Sequential Monte Carlo Methods for Dynamic Data Driven Simulation. 10.57709/5817785
  • Matthew O’Kelly et al. (2018). Scalable End-to-End Autonomous Vehicle Testing via Rare-event Simulation. 10.48550/arxiv.1811.00145

accelerated / statistical / event

Accelerated and certifiable evaluation of automated vehicles: Kriging and kernel surrogates for on-track testing, joint-distribution background-traffic models, and deep probabilistic accelerated evaluation with robust rare-event guarantees for black-box systems.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Mansur Arief et al. (2026). Certifiable Deep Importance Sampling for Rare-Event Simulation of Black Box Systems. 10.1287/opre.2021.0699
  • Mansur Arief et al. (2020). Deep Probabilistic Accelerated Evaluation: A Robust Certifiable Rare-Event Simulation Methodology for Black-Box Safety-Critical Systems. 10.48550/arxiv.2006.15722
  • Zhiyuan Huang (2020). Data-Driven Methods and Applications for Optimization under Uncertainty and Rare-Event Simulation. W3092550705
  • Zhiyuan Huang et al. (2018). A Versatile Approach to Evaluating and Testing Automated Vehicles based on Kernel Methods. 10.23919/acc.2018.8431590
  • Zhiyuan Huang et al. (2017). Towards affordable on-track testing for autonomous vehicle — A Kriging-based statistical approach. 10.1109/itsc.2017.8317656
  • Zhiyuan Huang et al. (2017). Towards Affordable On-track Testing for Autonomous Vehicle - A Kriging-based Statistical Approach. 10.48550/arxiv.1707.04897

sampling / importance / failure

Sampling-based estimation and control for autonomy: importance sampling for black-box AV failure probabilities (reachability-guided, state-dependent, diffusion-based, exponential tilting), plus sampling-based MPC (MPPI variants), POMDP planning and Monte Carlo localization reliability.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Tanmay Khandait et al. (2026). Diffusion-Guided Search via Exponential Tilting (DiffTilt): An Application to Falsification of Safety-Critical Systems. 10.48550/arxiv.2607.23134
  • Elizabeth Dietrich et al. (2026). Importance Sampling for Statistical Certification of Viable Initial Sets. 10.24433/co.0191384.v1
  • Harrison Delecki et al. (2024). Failure Probability Estimation for Black-Box Autonomous Systems using State-Dependent Importance Sampling Proposals. 10.48550/arxiv.2412.02154
  • Harrison Delecki et al. (2024). Diffusion-Based Failure Sampling for Evaluating Safety-Critical Autonomous Systems. 10.48550/arxiv.2406.14761
  • Elia Trevisan et al. (2024). Biased-MPPI: Informing Sampling-Based Model Predictive Control by Fusing Ancillary Controllers. 10.1109/lra.2024.3397083
  • Bianca Bendris (2019). Decentralized Stochastic Optimal Control for a Swarm of Micro Aerial Vehicles. W3136255470

dynamical / equilibrium / mechanics

Theory of rare transitions in dynamical systems outside driving: large-deviation theory in climate and fluid dynamics, variational path sampling, noise-driven transitions in non-equilibrium systems, AMOC resilience and ballistic-capture sets for spacecraft.

• 代表論文(新しい順 + いちばん古い 1 本):
  • Aditya Narayan Singh et al. (2025). Variational Path Sampling of Rare Dynamical Events. 10.1146/annurev-physchem-083122-115001
  • Aditya Narayan Singh et al. (2025). Variational path sampling of rare dynamical events. 10.48550/arxiv.2502.01852
  • Mickaël D. Chekroun et al. (2025). Non-Markovian reduced models to unravel transitions in non-equilibrium systems. 10.1088/1751-8121/ada7ad
  • Valérian Jacques-Dumas et al. (2024). Resilience of the Atlantic Meridional Overturning Circulation. 10.48550/arxiv.2407.04740
  • Gianmario Merisio (2022). EXTREMA: Ballistic capture sets at Mars with initial epoch December 9, 2023. 10.5281/zenodo.6103720
  • Vera Melinda Gálfi et al. (2021). Applications of large deviation theory in geophysical fluid dynamics and climate science. 10.1007/s40766-021-00020-z

piecewise / mixture / distribution

Piecewise mixture models for accelerated evaluation of automated vehicles, centred on the frontal cut-in scenario built from naturalistic driving data.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Zhiyuan Huang et al. (2017). Accelerated Evaluation of Automated Vehicles Using Piecewise Mixture Models. 10.1109/tits.2017.2766172
  • Zhiyuan Huang et al. (2017). Evaluation of automated vehicles in the frontal cut-in scenario — An enhanced approach using piecewise mixture models. 10.1109/icra.2017.7989024
  • Zhiyuan Huang et al. (2017). Accelerated Evaluation of Automated Vehicles Using Piecewise Mixture Models. 10.48550/arxiv.1701.08915
  • Zhiyuan Huang et al. (2016). Evaluation of Automated Vehicles in the Frontal Cut-in Scenario - an Enhanced Approach using Piecewise Mixture Models. 10.48550/arxiv.1610.09450
  • Zhiyuan Huang et al. (2016). Using the Piecewise Mixture Model to Evaluate Automated Vehicles in the Frontal Cut-in Scenario. W2592619158

sampling / rare / importance

Importance sampling and splitting refined with learned models: normalizing-flow latent-space proposals, temporal importance sampling for AV validation, adaptive splitting of reusable temporal monitors, low-rank mixture proposals in high dimension, and IS-accelerated fault injection for DL accelerators.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Liam A. Kruse et al. (2025). Scalable Importance Sampling in High Dimensions with Low-Rank Mixture Proposals. 10.48550/arxiv.2505.13335
  • Liam A. Kruse et al. (2025). Enhanced Importance Sampling Through Latent Space Exploration in Normalizing Flows. 10.1609/aaai.v39i17.33978
  • Liam A. Kruse et al. (2025). Enhanced Importance Sampling through Latent Space Exploration in Normalizing Flows. 10.48550/arxiv.2501.03394
  • Sandeep Thalapanane et al. (2024). TRAVERSE: Traffic-Responsive Autonomous Vehicle Experience & Rare-event Simulation for Enhanced safety. 10.1109/iros58592.2024.10802298
  • Craig Innes et al. (2024). Adaptive Splitting of Reusable Temporal Monitors for Rare Traffic Violations. 10.1109/iros58592.2024.10802747
  • Sirisha Rambhatla et al. (2022). I-SEA: Importance Sampling and Expected Alignment-Based Deep Distance Metric Learning for Time Series Analysis and Embedding. 10.1609/aaai.v36i7.20776

traffic / simulation / vehicle

114 本

Traffic and vehicle simulation platforms for automated driving: open simulators (CARLA, LGSVL, AWSIM, AutoDRIVE), co-simulation with SUMO/VISSIM and vehicle dynamics, V2X and cybersecurity testbeds, mixed CAV traffic studies and driver-in-the-loop VR simulators.

• Co-simulation coupling: Synchronising microscopic traffic (SUMO/VISSIM) with a 3D simulator and vehicle dynamics is the standard architecture for large-scale AV testing.
• Distributed simulation: Multi-GPU or multi-host CARLA/Autoware setups scale multi-vehicle CAV and valet-parking scenarios.
• V2X modelling: Message generation from OpenSCENARIO, network simulators and attack injection evaluate cooperative driving.
• Mixed-traffic impact metrics: Throughput, string stability, energy use and safety surrogates quantify CAV penetration effects.
• Sensor-model validation: Radar, LiDAR noise, RTK GNSS and IMU fidelity are checked against real data before results are trusted.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• 交通流と他車の振る舞い: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、social_force_step(歩行者の Social Force、points × points × points → any)
• 自前の世界(真値つきの描画とセンサー): world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、gs_from_world(世界 → 3DGS、table → table)、carla_scene_world(CARLA の場面を自前の世界に(同じ場面を両方で)、table → table)、carla_depth_decode(CARLA の深度画像(RGB 符号)を m に、rgb → image2d)、carla_transform_matrix(CARLA の座標系(左手・度)↔ 自前の行列、any → matrix)、cr_read(CommonRoad のシナリオ XML を読む、any → table)、cr_drive(CommonRoad のシナリオを運転して採点、table → table)、osm_road_graph(OSM の道路網から地図、table → table)
• 通信(V2X・可視光通信) —— op なし(Fullseye の外)

distributed / traffic / management

Distributed and co-simulation architectures around CARLA, AWSIM and Autoware: multi-GPU CARLA for multi-agent CAVs, distributed dynamic-map generation, CARLA-VISSIM corridors, elevation-aware 2D/3D traffic coupling and human-in-the-loop frameworks.

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 1 個: social_force_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • — (2026). DISTRIBUTED CARLA ARCHITECTURE FOR MULTI-AGENT CAV SIMULATION: DESIGN AND PRELIMINARY MULTI-GPU EVALUATION. 10.17683/ijomam/issue25/v2.3
  • Patrick Rebling et al. (2026). CARLAverse: A Highly Modular, Distributed, and Multimodal Framework for Human-in-the-Loop Simulation. 10.48550/arxiv.2609.11478
  • Diego Caballero García-Alcaide et al. (2026). Gestión inteligente de semáforos mediante visión y aprendizaje por refuerzo. 10.17979/ja-cea.2026.47.13792
  • Sima Ashayer et al. (2026). A Corridor-Scale CARLA-VISSIM Co-Simulation Framework for Multi-Intersection Urban Traffic. 10.48550/arxiv.2606.15431
  • Zubair Islam et al. (2026). DMV-AVP: Distributed Multi-Vehicle Autonomous Valet Parking Using Autoware. 10.48550/arxiv.2601.16327
  • Anton Debner et al. (2019). Scalability of a Machine Learning Environment for Autonomous Driving Research. 10.1109/indin41052.2019.8972278

interaction / simulators / automation

Driving simulators for human behaviour and interaction research: VR extensions of CARLA and AWSIM for drivers and pedestrians, SUMO plus Unreal bespoke simulators, XR replay of real drive data, haptic ADAS cues and BIM-derived virtual environments.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Junxi Zhang (2026). BIM Technology in Intelligent Driving Simulation: Potentials and Challenges to Address Virtual Environment Deficiencies. 10.54254/2755-2721/2026.ba32187
  • Arfa Bhandari (2025). AutoNav: A CARLA-Based Framework for Autonomous Vehicle Simulation. 10.26562/ijirae.2025.v1210.09
  • Jason Sutanto et al. (2025). Automating the Future of Inland Shipping: Photorealistic and Physically Accurate Simulations for Inland Waterway Transport. 10.1088/1742-6596/3123/1/012060
  • Renan Florencio et al. (2025). Fooling the Model, Failing the Road: Benchmarking Inertial Sensor Fidelity in Driving Simulators. 10.5753/eniac.2025.12416
  • Leon Sebastian Fuessner et al. (2025). Interactive Visualization of Real-World Automated Driving Data using AWSIM and VARJO-XR4. 10.1145/3744335.3758969
  • Satu Innamaa et al. (2018). Framework for assessing the impacts of automated driving. 10.5281/zenodo.1485282

sumo / figshare / codrive

Data and software deposits for co-simulation platforms: OpenCAMS (traffic, 3D driving and network co-simulation), E-CoDrive for energy-aware scenario testing of electric AVs, a beyond-5G V2X cybersecurity test environment and adversarial attack generation material.

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 1 個: social_force_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • CHRISTOS ANAGNOSTOPOULOS et al. (2026). Supplementary Videos for “A Unified Interactive Simulation Platform for Smart Mobility: Adversarial Attack Generation in Connected Autonomous Vehicle”. 10.6084/m9.figshare.30951002
  • CHRISTOS ANAGNOSTOPOULOS et al. (2026). Supplementary Videos for “A Unified Interactive Simulation Platform for Smart Mobility: Adversarial Attack Generation in Connected Autonomous Vehicle”. 10.6084/m9.figshare.30951002.v1
  • Napolitano et al. (2026). E-CoDrive. 10.6084/m9.figshare.32244783
  • Napolitano et al. (2026). E-CoDrive. 10.6084/m9.figshare.32244783.v1
  • João Paulo Javidi da Costa (2025). B5GCyberTestV2X - Beyond 5G Virtuelle Umgebung für Cybersicherheitstests von V2X-Systemen. 10.34657/34387
  • M. B. Ahmad et al. (2025). OpenCAMS: An Open-Source Connected and Automated Mobility Co-Simulation Platform for Advancing Next-Generation Intelligent Transportation Systems Research. 10.48550/arxiv.2507.09186

traffic / vehicle / simulation

Traffic simulation for AV testing: naturalistic and distributionally consistent background traffic, safety-critical scenario generation from crash data and VLMs, game-theoretic interaction models, simulator reviews, re-simulation analysis and high-fidelity 4D closed-loop worlds.

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 14 個: social_force_step、world_build、world_camera、world_move、lidar_scan、env_render、tile_stream、gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Tara Rajabi Nezhad et al. (2026). Driving Simulators for Autonomous Vehicles: Comprehensive Review of Current Applications and Research Trends. 10.3390/vehicles8070158
  • Hongyi Zhao et al. (2026). SG-CADVLM: a context-aware decoding powered vision-language model for safety-critical scenario generation in autonomous driving. 10.1080/23249935.2026.2666104
  • Nahian Salsabil et al. (2026). TRACE: Topology-aware Reconstruction of Accidents in CARLA for AV Evaluation. W7157506450
  • Hongyi Zhao et al. (2026). SG-CADVLM: A Context-Aware Decoding Powered Vision Language Model for Safety-Critical Scenario Generation. 10.48550/arxiv.2601.18442
  • Xiaoyu Zhang (2025). Advancing Safety-Critical Scenario Generation for Autonomous Vehicle Testing: Integrating In-Depth Crash Data and Advanced Generative Models. 10.33774/coe-2025-rjqw5
  • Jincai Chang et al. (2017). Modeling and simulations on automated vehicles to alleviate traffic congestion. 10.21595/mme.2017.19302

v2x / communication / vehicular

V2X communication and cybersecurity in simulation: network-plus-traffic co-simulation, attack risk on V2V platoons, I2V communication in truck-platooning scenarios, OpenSCENARIO-driven V2X message generation, and controllable large-scale traffic simulators (LimSim, LCSim).

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 1 個: social_force_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • Minhaj Uddin Ahmad et al. (2026). End-to-End Co-Simulation Testbed for Cybersecurity Research and Development in Intelligent Transportation Systems. 10.1142/9789819831548_0006
  • Hanan H. Hussein et al. (2026). V2X for smart mobility: communication protocols, localization techniques and simulation frameworks. 10.1186/s43067-026-00362-w
  • Marcelo Alves et al. (2025). Parking State Monitoring via Ultrasonic Sensors: Large Scale Co-Simulation Validation. 10.1109/icufn65838.2025.11169971
  • Yuheng Zhang et al. (2024). LCSim: A Large-Scale Controllable Traffic Simulator. 10.48550/arxiv.2406.19781
  • Kevin Herman Muraro Gularte et al. (2024). Integrating Cybersecurity in V2X: A Review of Simulation Environments. 10.1109/access.2024.3504404
  • Katharina Hartmann (2020). Konzeption und Entwicklung eines Moduls für die Generierung von V2X Nachrichten aus OpenSCENARIO Daten als Teil einer HiL Laborumgebung. W7127548747

cav / cavs / traffic

Connected and automated vehicles in mixed traffic: integrated traffic/vehicle co-simulation, platoon effects on flow, energy efficiency, adverse-weather impacts, cooperative perception and RSU-assisted risk assessment, plus the winning modular stack of CARLA Leaderboard 2.0.

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 1 個: social_force_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • Shunsuke Ito et al. (2026). VAST: V2X/Dynamic Map-Aware Autonomous Driving Systems Validation Toolchain. W7213675658
  • Luca Forneris (2026). Closing the Loop in Automated Driving: Dataset Design, Scenario Synthesis, and Deployed Perception. 10.15167/forneris-luca_phd2026-04-24
  • Dingqian Cui et al. (2026). Influence of PID controlled CAV platoon on mixed traffic flow. 10.1117/12.3095962
  • Nikola Topić et al. (2025). Co-simulation Supported Multicriteria Analysis of Autonomous Vehicles Impact on Urban Motorway Traffic Flow. 10.1016/j.trpro.2025.10.062
  • Muamer Abuzwidah et al. (2024). Assessing the Impact of Adverse Weather on Performance and Safety of Connected and Autonomous Vehicles. 10.28991/cej-2024-010-09-019
  • Jakob Kaths et al. (2017). Integration of DYNA4 Car Professional with SUMO to perform virtual testing of automated vehicles in complex surrounding traffic. W2749713749

traffic / sumo / road

Small leaf on traffic-level tools: a CARLA Town05 network adapted for SUMO co-simulation via OpenDRIVE, vision-based signal control with microscopic simulation, portable trajectory prediction, steering-angle testing and a naturalistic dataset recorded during blackouts.

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 1 個: social_force_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • Angela Carboni et al. (2026). Town05 road network modified and ready to use for traffic simulation in SUMO and co-simulation between SUMO and CARLA. 10.17632/xhbwtky3sz.1
  • Angela Carboni et al. (2026). Town05 road network modified and ready to use for traffic simulation in SUMO and co-simulation between SUMO and CARLA. 10.17632/xhbwtky3sz
  • Supriya Sarker et al. (2024). Beacon: A Naturalistic Driving Dataset During Blackouts for Benchmarking Traffic Reconstruction and Control. 10.48550/arxiv.2412.14208
  • Borna Benja et al. (2024). Steering Angle Prediction and Algorithm Performance Testing in Autonomous Driving Simulators. 10.1109/elmar62909.2024.10694075
  • Pan He et al. (2024). A Simple Framework Towards Vision-based Traffic Signal Control with Microscopic Simulation. 10.48550/arxiv.2403.06884
  • Cheng Wei et al. (2023). An Easy-portable Displacement-offset-based Trajectory Prediction Method for Autonomous Vehicles. 10.1109/dsa59317.2023.00088

simulator / simulation / autoware

Core autonomous-driving simulators and comparisons: CARLA, LGSVL, AutoDRIVE, CARLA-Autoware bridges, customized co-simulation and reviews contrasting CARLA, AirSim and Gazebo.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Griffen Agnello et al. (2026). Toward Robust Autonomous Driving in Rural and Off-Road Environments: A Perception Based Approach. 10.1109/most69733.2026.00023
  • Xuanyu Zhou (2026). A Comparative Review of Autonomous Driving Simulation Platforms: Carla, AirSim, and Gazebo. 10.54254/2755-2721/2026.bj32062
  • Yan Ai et al. (2025). Reproduction of Real-World Scenarios in CARLA: An Extension of CARLA Functionality. 10.4271/2025-01-8059
  • Marcondes et al. (2025). Métodos de reconstrução tridimensional de cenários de condução veicular para assistência ao condutor. W7120486564
  • Ziwei Song et al. (2024). Autoware.Flex: Human-Instructed Dynamically Reconfigurable Autonomous Driving Systems. 10.48550/arxiv.2412.16265
  • Alexey Dosovitskiy et al. (2017). CARLA: An Open Urban Driving Simulator. 10.48550/arxiv.1711.03938

scenario / testing / automated

210 本

Scenario-based testing and safety validation of automated driving systems: scenario description languages (OpenSCENARIO, CommonRoad), scenario databases and libraries, criticality metrics, safety-critical scenario generation, accelerated evaluation and test frameworks for CARLA, Apollo and Autoware. Many papers address homologation questions such as coverage, pass/fail criteria and simulation credibility.

• Functional-logical-concrete hierarchy: Scenarios are refined from prose to parameter ranges to concrete instances; testing samples the logical parameter space.
• Criticality metrics: TTC, PET, required deceleration, acceleration-based and model-predictive metrics rate how dangerous a scenario is.
• Scenario formats: OpenSCENARIO storyboards (Init, Story, Act, ManeuverGroup, Event, Action, Condition) with OpenDRIVE roads; converters to CommonRoad exist.
• Coverage and diversity: Trajectory clustering, scenario similarity measures and ODD ontologies decide whether the test set is complete enough.
• Simulation credibility: Flaky tests, game-engine nondeterminism and simulator ensembles affect whether a virtual verdict can be trusted.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• シナリオ試験と危険度の指標: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、turn_maneuver_check(右左折の採点、table → table)、skill_test_score(技能試験の減点、table → table)
• 稀な事象の加速評価(重要度サンプリング): importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)
• 自前の世界(真値つきの描画とセンサー): world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、gs_from_world(世界 → 3DGS、table → table)、carla_scene_world(CARLA の場面を自前の世界に(同じ場面を両方で)、table → table)、carla_depth_decode(CARLA の深度画像(RGB 符号)を m に、rgb → image2d)、carla_transform_matrix(CARLA の座標系(左手・度)↔ 自前の行列、any → matrix)、cr_read(CommonRoad のシナリオ XML を読む、any → table)、cr_drive(CommonRoad のシナリオを運転して採点、table → table)、osm_road_graph(OSM の道路網から地図、table → table)

scenario / scenarios / criticality

Identifying and generating safety-critical scenarios: reviews of criticality metrics, scenario-in-the-loop identification concepts, causal-influence and latent-flow diversified generation, diffusion-based generation (DiffScene), NLP-derived scenarios from disengagement reports and near-crash detection at roundabouts.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Chejian Xu et al. (2025). DiffScene: Diffusion-Based Safety-Critical Scenario Generation for Autonomous Vehicles. 10.1609/aaai.v39i8.32951
  • Dingcheng Gao et al. (2024). Diversifying Latent Flows for Safety-Critical Scenarios Generation. 10.1109/itsc58415.2024.10919760
  • Yibing Yang et al. (2024). Safety-Critical Scenario Generation by Causal Influence Detection. 10.1109/itsc58415.2024.10920225
  • Qunying Song et al. (2024). Generating Executable Test Scenarios from Autonomous Vehicle Disengagements using Natural Language Processing. 10.1145/3643915.3644098
  • Juan Trullos et al. (2023). Criticality dimension-based probabilistic framework to detect near crashes in a roundabout. 10.1186/s12544-023-00602-4
  • Shuo Feng et al. (2020). Testing Scenario Library Generation for Connected and Automated Vehicles, Part I: Methodology. 10.1109/tits.2020.2972211

scenario / testing / test

Methodology and trustworthiness of scenario-based ADS testing: simulator ensembles, statistical confidence in safety claims, combinatorial test construction, OpenSCENARIO fragment datasets, flaky tests in simulators, trajectory clustering for diversity, coverage questions, pass/fail criteria and corner-case datasets from naturalistic data.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Zhuoren Li et al. (2026). Comparison-Based Ordinal Learning for Proactive Driving Risk Assessment. 10.48550/arxiv.2607.11128
  • Thoshitha Gamage et al. (2026). OSC2Runner: OpenSCENARIO 2.x Compliant High-Fidelity AV Simulation in CARLA. 10.48550/arxiv.2606.26533
  • Thoshitha Gamage et al. (2026). Compiling OpenSCENARIO 2.1 for Scenario-Based Testing in CARLA. 10.48550/arxiv.2604.16452
  • Lev Sorokin et al. (2026). Simulator ensembles for trustworthy autonomous driving systems testing. 10.1007/s10664-026-10821-7
  • Sungmo Ku et al. (2026). Rule-Based Scenario Classification Using Vehicle Trajectories. 10.3390/ijgi15010037
  • Walther Wachenfeld (2017). How Stochastic can Help to Introduce Automated Driving. 10.26083/tuprints-00005949

testing / survey / software

Surveys and frameworks for ADS validation and black-box safety testing: taxonomies of validation approaches, scenario-generation surveys, formal scenario-based testing (Scenic/VerifAI style), interpretable failure search, game-engine determinism, causal-inference testing, safety shields and LLM use in scenario testing.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Michael Foster et al. (2025). Using Causal Inference to Test Systems with Hidden and Interacting Variables: An Evaluative Case Study. 10.1145/3756681.3756967
  • Yongqi Zhao et al. (2025). A Survey on the Application of Large Language Models in Scenario-Based Testing of Automated Driving Systems. 10.48550/arxiv.2505.16587
  • Michael S. Foster et al. (2025). Using Causal Inference to Test Systems with Hidden and Interacting Variables: An Evaluative Case Study. 10.48550/arxiv.2504.16526
  • Ziyu Wang et al. (2024). A Survey of Scenario Generation for Automated Vehicle Testing and Validation. 10.3390/fi16120480
  • Filip Cano et al. (2023). Safety Shielding under Delayed Observation. 10.1609/icaps.v33i1.27181
  • Cumhur Erkan Tuncali et al. (2019). Requirements-Driven Test Generation for Autonomous Vehicles With Machine Learning Components. 10.1109/tiv.2019.2955903

testing / scenario / ads

Virtual test platforms and processes for automated vehicles: X-in-the-loop validation, OpenSCENARIO in dSPACE, open-source scalable virtual test environments, driver-vehicle and agent models in open simulation architectures, ODD ontologies, GIDAS-based criticality analysis, model-predictive safety metrics and diversity-oriented exploration.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Christian Geller et al. (2026). Scalable Distributed Simulation-Based Testing for Automated Driving Systems. 10.48550/arxiv.2608.20904
  • Junze Sun et al. (2026). Deep Reinforcement Learning Based Safety-critical Scenario Generation for Autonomous Ship Digital Testing. 10.1088/1742-6596/3299/1/012009
  • Qunying Song et al. (2026). In the Driver’s Seat: A Multi-Company Study on the Reality of Autonomous Driving System Testing. 10.48550/arxiv.2607.15820
  • Leticia Fernández Moguel et al. (2026). From ODDs to simulation : automatic railway scenario generation. 10.21256/zhaw-37046
  • Christian Geller et al. (2026). Integration of an agent model into an open simulation architecture for scenario-based testing of automated vehicles. 10.1515/auto-2025-0066
  • Florian Bock Christoph Sippl et al. (2016). From Simulation Data to Test Cases for Fully Automated Driving and ADAS - A Multilayer Model Concept. W2746672140

scenario / automated / criticality

The largest leaf on scenario-based safeguarding: test concepts, logical-scenario evaluation, CommonRoad Scenario Factory, functional decomposition against parameter explosion, collision-criticality metrics, ODD-based generation, OpenSCENARIO generation, scenario similarity and vocabulary formalisation, and perception-testing reviews.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Jörg Gamerdinger et al. (2026). Scenario-Independent Criticality Assessment and Prediction for Vulnerable Road Users in Autonomous Driving. 10.48550/arxiv.2609.11947
  • Olek Osikowicz et al. (2026). Multi-Fidelity Bayesian Optimization for Simulation Based Autonomous Driving Systems Testing. 10.1109/iv66570.2026.11623850
  • Mingfei Cheng et al. (2026). Drivora: A Unified and Extensible Infrastructure for Search-based Autonomous Driving Testing. 10.1145/3774748.3787599
  • Lutz Eckstein et al. (2026). Test Concepts for the Safeguarding of Automated Driving. 10.1007/978-3-658-45276-6_46
  • Stefan Jesenski (2026). Simulation-based Methods for the Validation of Highly Automated Driving. 10.5445/ir/1000189911
  • Philipp Junietz et al. (2018). Criticality Metric for the Safety Validation of Automated Driving using Model Predictive Trajectory Optimization. 10.1109/itsc.2018.8569326

test / accelerated / vehicles

Test acceleration and scenario libraries from naturalistic data: highD drone trajectories, accident-derived cut-in generalisation, naturalistic-and-adversarial environments for driving-intelligence tests, importance sampling for lane-change and car-following, subset simulation, OpenSCENARIO generation for ADAS, closed-loop fuzzing and simulator comparisons.

• 使う op: importance_risk_estimate(発生率を上げて走らせ、重みで本来の事故率に戻す(不偏)、any → table)、poisson_events(稀な事象を頻度つきで置く、any → signal)、poisson_events_xt(場所と時刻で変わる発生率 λ(x, t)、any → any)、bus_stop_rate(停車中のバスの前後で上がる飛び出しの率、any → any)、ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、ほか 6 個: rss_longitudinal_check、rss_lateral_check、dilemma_zone、occlusion_safe_speed、turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Jiawan Ye et al. (2025). GE-Div: A Gaussian Entropy-Based Metric for Trajectory Prediction Diversity Assessment. 10.1109/aiotc66747.2025.11198708
  • Yuening Hu et al. (2025). A simulation-based testing framework for autonomous driving: ensuring realism and priority of test scenarios. 10.1080/15472450.2025.2497510
  • Shu Li et al. (2024). Few-Shot Testing of Autonomous Vehicles with Scenario Similarity Learning. 10.48550/arxiv.2409.14369
  • Pingfei Li et al. (2024). Generalization of cut-in pre-crash scenarios for autonomous vehicles based on accident data. 10.1038/s41598-024-68263-4
  • Arpit S Agarkar et al. (2023). Simulation based Closed Loop Scenario Fuzzing for Safety Critical ADAS Applications. 10.1109/indicon59947.2023.10440801
  • Ding Xuan Zhao et al. (2015). Accelerated Evaluation of Automated Vehicles in Lane Change Scenarios. 10.1115/dscc2015-9718

testing / software / virtual

Testing frameworks and tools for AV stacks: ViSTA, scenario-driven development for Autoware, roadside-object fuzzing, OpenSCENARIO-to-CommonRoad conversion, credibility assessment, perception error models, fault tolerance of multi-sensor fusion, CARLA-Apollo co-simulation fuzzing, CARLA hardware-in-the-loop and violation cause analysis.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 15 個: turn_maneuver_check、skill_test_score、world_build、world_camera、world_move、lidar_scan、env_render、tile_stream、gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Xiaodong Zhang et al. (2026). CapCo: Automating Carla-Apollo Co-Simulation and Scenario Fuzzing. 10.1145/3803437.3806401
  • Yang Sun et al. (2026). Natural Adversaries: Fuzzing Autonomous Vehicles with Realistic Roadside Object Placements. 10.1109/icst69053.2026.00019
  • Gehrke et al. (2025). vEDGAR – Can CARLA Do HiL?. 10.48550/arxiv.2512.08541
  • Arif Hossain (2025). Study on Data-Driven Scenario Construction for Autonomous Driving Testing. 10.33774/coe-2025-jz51g
  • Yukuan Yang et al. (2025). AuthSim: Toward Authentic and Effective Safety-Critical Scenario Generation for Autonomous Driving Tests. 10.1109/tits.2025.3586608
  • Yunhan Jia et al. (2017). Towards secure and safe appified automated vehicles. 10.1109/ivs.2017.7995800

scenario / testing / automated

Foundations and infrastructure of scenario-based testing: surveys, paradigms and fundamental considerations for homologation, clustering of scenario space, scenario description languages, reproducible scenario-database architectures, VR authoring, declarative scenarios (RoadLogic), PNT-aware scenarios and the Safety2Drive benchmark.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Patrick Irvine et al. (2026). Enabling PNT-Aware scenario-based testing of automated driving systems. W7164273645
  • Yavar Taghipour Azar et al. (2026). A Reproducible Reference Architecture for Automated Driving Scenario Databases. 10.3390/vehicles8040088
  • Ezio Bartocci et al. (2026). Declarative Scenario-based Testing with RoadLogic. 10.48550/arxiv.2603.09455
  • Naya Baslan et al. (2025). Safety and Reliability: Validation for Automated Driving Functions through Scenario-Based Testing. 10.1109/iv64158.2025.11097692
  • J. Jenny Li et al. (2025). Safety2Drive: Safety-Critical Scenario Benchmark for the Evaluation of Autonomous Driving. 10.48550/arxiv.2505.13872
  • Stefan Riedmaier et al. (2020). Survey on Scenario-Based Safety Assessment of Automated Vehicles. 10.1109/access.2020.2993730

weather / detection / synthetic

178 本

Perception under adverse weather and the synthetic-to-real gap: physics-based fog, rain and snow models for LiDAR and camera, radar simulation and camera-radar-LiDAR fusion in bad weather, weather datasets and test facilities, synthetic data generation and domain adaptation for detection and segmentation.

• LiDAR weather physics: Beer-Lambert attenuation plus backscatter from droplets yields fewer returns, shorter range and spurious near-range points; LISA-style augmentation applies Mie-based scattering.
• Weather test facilities: Controlled fog/rain chambers (e.g. CEREMA, PAVIN) and their digital twins provide ground truth for validating weather models.
• Radar in bad weather: Radar degrades least in fog and snow; datasets such as RADIATE and data-driven radar models support simulation.
• Domain adaptation: Style transfer, curriculum and self-training adaptation reduce synthetic-to-real accuracy loss; measure the gap by train-on-sim/test-on-real mAP or mIoU.
• Camera effect modelling: Adding blur, noise, chromatic aberration and exposure effects makes synthetic images train better models.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• 天候と視程(閉形式で根拠づける): koschmieder(視程と消散係数(Koschmieder)、any × any × scalar × any → any)、beta_from_mor(気象光学距離 → 消散係数、scalar → scalar)、fog_beta_from_profile(画像の明るさの縦断から霧の濃さ、any × any × scalar × scalar × scalar → table)、veiling_luminance(まぶしさの光幕輝度、any × any → any)、sight_stop_speed(見える距離で止まれる速さ、scalar × scalar × scalar → scalar)、env_params(天候の母数、入力なし → table)
• sim-to-real の比較(同じ場面を対で測る): psnr(画素の差、image2d × image2d → scalar)、ssim(構造の類似、image2d × image2d → scalar)、ms_ssim(多段の構造類似、image2d × image2d → scalar)、ssim_map(どこが違うかの地図、image2d × image2d → image2d)、fsim(位相一致と勾配の知覚的類似(FSIM)、image2d × image2d → scalar)、gmsd(勾配の類似の偏差(GMSD)、image2d × image2d → scalar)、vifp(情報忠実度(VIFp)、image2d × image2d → scalar)、tid2013_evaluate(指標を人の評価(TID2013)と順位相関で採点、any × any → table)、world_camera(真値つきの対の描画、table × matrix × matrix → table)
• LiDAR(距離・強度・幾何): lidar_spec(走査の仕様、入力なし → table)、lidar_scan(世界を走査、mesh × table × matrix → table)、ray_plane_range(平面までの距離(閉形式)、points × points → signal)、ray_box_ranges(箱までの距離(閉形式)、points × points → signal)、project_spherical(球面投影、points → image2d)、tof_multipath_phase(凹みの多重反射で ToF の位相がずれる量(閉形式)、入力なし → table)、tof_transport_depth(輸送行列で多重反射を入れた ToF の深度、table → table)
• レーダー・ドップラー・音の到来方向: fmcw_range_profile(FMCW の距離、beatcube → signal)、range_doppler_map(距離・速度の地図、beatcube → image2d)、doppler_shift(ドップラーの閉形式、any → any)、doppler_track(サイレンの周波数の追跡、signal → table)、tdoa_bearing(到着時間差から方向、signal × signal → table)

radar / fusion / sensor

Radar and multi-sensor fusion for all-weather perception: Monte Carlo FMCW LiDAR-radar fusion in fog, deep fusion in unseen adverse weather, the RADIATE bad-weather radar dataset, realistic radar simulators for CARLA, data-driven radar models (Bayesian Gaussian mixtures), roadside radar-vision fusion simulators and 4D radar perception reviews.

• 使う op: fmcw_range_profile(FMCW の距離、beatcube → signal)、range_doppler_map(距離・速度の地図、beatcube → image2d)、doppler_shift(ドップラーの閉形式、any → any)、doppler_track(サイレンの周波数の追跡、signal → table)、tdoa_bearing(到着時間差から方向、signal × signal → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Leila Mansour (2026). WeatherPrompt-Fusion: Prompt-Guided Multi-Modal Perception for Autonomous Driving in Adverse Weather. 10.54097/29prnw95
  • Ana Torres et al. (2026). UA-CF: Uncertainty-Aware Camera-LiDAR Fusion for Robust 3D Object Detection Under Adverse Weather. 10.54097/ybd75q69
  • Zhixiang Xu (2026). Multi-Sensor Fusion Strategies for Robust Autonomous Driving Perception under Adverse Weather and Complex Urban Traffic Conditions. 10.1051/matecconf/202642402001
  • Hong Xijie (2026). Optimization Research on Autonomous Driving Environmental Perception Algorithm for Complex Weather (Heavy Rain/Heavy Snow/Fog). 10.47297/taposatwsp2633-456912.20260704
  • Satyam Srivastava et al. (2025). A Realistic Radar Simulator for End-to-End Autonomous Driving in CARLA. 10.1109/vtc2025-fall65116.2025.11310463
  • Rui Song et al. (2019). A Multi-Sensor Simulation Environment for Autonomous Cars. 10.23919/fusion43075.2019.9011278

synthetic / domain / detection

Training and validating perception with synthetic data: automated synthetic dataset generation for 3D detection, simulator-generated data for AV perception, geometric and curriculum domain adaptation for segmentation, LiDAR noise models for sim-to-real, instance-level safety-aware fidelity and optimal sim training strategies.

• 使う op: psnr(画素の差、image2d × image2d → scalar)、ssim(構造の類似、image2d × image2d → scalar)、ms_ssim(多段の構造類似、image2d × image2d → scalar)、ssim_map(どこが違うかの地図、image2d × image2d → image2d)、fsim(位相一致と勾配の知覚的類似(FSIM)、image2d × image2d → scalar)、gmsd(勾配の類似の偏差(GMSD)、image2d × image2d → scalar)、ほか 3 個: vifp、tid2013_evaluate、world_camera
• 代表論文(新しい順 + いちばん古い 1 本):
  • G. Damico et al. (2026). SynDRA-Gen: A Configurable Pipeline for Controlled Synthetic Multi-Modal Data Generation in a Railway Shunting Yard. 10.4203/ccc.15.20.1
  • Tanel Liiv et al. (2026). Training with synthetic data for drone detection in thermal imagery. 10.48550/arxiv.2608.17799
  • Paul Schulz et al. (2025). Automating synthetic dataset generation for image-based 3D detection: a literature review. 10.1007/s10462-025-11431-3
  • Corentin Bunel (2025). Détection d’objets inconnus en monde ouvert pour les systèmes de conduite autonome. W7122348150
  • Kromm et al. (2025). Data Fusion for Scene Graph Generation: Bridging Simulated and Real-World Datasets. 10.34734/fzj-2025-03665
  • Yang Zhang et al. (2017). Curriculum Domain Adaptation for Semantic Segmentation of Urban Scenes. 10.1109/iccv.2017.223

domain / adaptation / detection

Domain gap and domain adaptation for driving perception: depth completion adaptation, continual adaptation for detection, diffusion style transfer, day-night segmentation, physics-based rendering realism versus randomisation, camera-effect modelling, LiDAR beam distillation and the CODA corner-case dataset.

• 使う op: psnr(画素の差、image2d × image2d → scalar)、ssim(構造の類似、image2d × image2d → scalar)、ms_ssim(多段の構造類似、image2d × image2d → scalar)、ssim_map(どこが違うかの地図、image2d × image2d → image2d)、fsim(位相一致と勾配の知覚的類似(FSIM)、image2d × image2d → scalar)、gmsd(勾配の類似の偏差(GMSD)、image2d × image2d → scalar)、ほか 3 個: vifp、tid2013_evaluate、world_camera
• 代表論文(新しい順 + いちばん古い 1 本):
  • Julian Truetsch et al. (2026). Understanding Autonomous Driving Datasets by Describing Differences between Image Subsets in Natural Language. 10.48550/arxiv.2609.03677
  • Estelle Chigot et al. (2025). Style transfer with diffusion models for synthetic-to-real domain adaptation. 10.1016/j.cviu.2025.104445
  • Chunzheng Li et al. (2025). FS$^{2}$D: Fully Sparse Few-Shot 3D Object Detection. 10.1109/lra.2025.3560868
  • Anh-Dzung Doan et al. (2023). Assessing domain gap for continual domain adaptation in object detection. 10.1016/j.cviu.2023.103885
  • Elysia Q. Smyers et al. (2023). AVOIDDS: Aircraft Vision-based Intruder Detection Dataset and Simulator. 10.48550/arxiv.2306.11203
  • Alexandra Carlson et al. (2018). Modeling Camera Effects to Improve Visual Learning from Synthetic Data. 10.48550/arxiv.1803.07721

synthetic / domain / real

Synthetic data across vision tasks, beyond driving: traversability, pose, tracking, depth, instance segmentation, raindrop removal, off-road and traffic-sign detection and remote-sensing disparity.

• 使う op: psnr(画素の差、image2d × image2d → scalar)、ssim(構造の類似、image2d × image2d → scalar)、ms_ssim(多段の構造類似、image2d × image2d → scalar)、ssim_map(どこが違うかの地図、image2d × image2d → image2d)、fsim(位相一致と勾配の知覚的類似(FSIM)、image2d × image2d → scalar)、gmsd(勾配の類似の偏差(GMSD)、image2d × image2d → scalar)、ほか 3 個: vifp、tid2013_evaluate、world_camera
• 代表論文(新しい順 + いちばん古い 1 本):
  • D.I. Ryabokon (2026). Synthetic Data Generation for Computer Vision Tasks in SITL Systems with Unreal Engine and Airsim. 10.15407/intechsys.2026.02.006
  • Asami Oganna et al. (2026). Dataset Creation and Alignment-Robust Learning for Real-World Raindrop Removal. 10.1109/iceic69189.2026.11386102
  • Frank Ruis et al. (2024). Improving Object Detector Training on Synthetic Data by Starting With a Strong Baseline Methodology. 10.48550/arxiv.2405.19822
  • Alexandra Duminil et al. (2024). A Comprehensive Exploration of Fidelity Quantification in Computer-Generated Images. 10.3390/s24082463
  • Giuseppe Vecchio et al. (2024). Terrain traversability prediction through self-supervised learning and unsupervised domain adaptation on synthetic data. 10.1007/s10514-024-10158-4
  • Adrien Gaidon et al. (2016). Virtual Worlds as Proxy for Multi-Object Tracking Analysis. 10.48550/arxiv.1605.06457

weather / adverse / detection

Adverse-weather perception and stress testing: GAN-simulated foggy LiDAR, robust detectors (YOLO variants, geometric constraints), image restoration and enhancement for driving, weather translation with diffusion and Pix2Pix, perception stress testing, style-based profiling of the synthetic-to-real gap, LiDAR spoofing in weather and lidar corner cases.

• 使う op: koschmieder(視程と消散係数(Koschmieder)、any × any × scalar × any → any)、beta_from_mor(気象光学距離 → 消散係数、scalar → scalar)、fog_beta_from_profile(画像の明るさの縦断から霧の濃さ、any × any × scalar × scalar × scalar → table)、veiling_luminance(まぶしさの光幕輝度、any × any → any)、sight_stop_speed(見える距離で止まれる速さ、scalar × scalar × scalar → scalar)、env_params(天候の母数、入力なし → table)、ほか 9 個: psnr、ssim、ms_ssim、ssim_map、fsim、gmsd、vifp、tid2013_evaluate、world_camera
• 代表論文(新しい順 + いちばん古い 1 本):
  • Viktoria Afxentiou (2026). Deep Learning for Efficient Adverse Weather Image Processing in Autonomous Driving. 10.25392/leicester.data.32947304
  • Viktoria Afxentiou (2026). Deep Learning for Efficient Adverse Weather Image Processing in Autonomous Driving. 10.25392/leicester.data.32947304.v1
  • Dingyi Yao et al. (2026). A Style-Based Profiling Framework for Quantifying the Synthetic-to-Real Gap in Autonomous Driving Datasets. 10.1109/iv66570.2026.11623866
  • Hamed Khatounabadi et al. (2026). WILD SAM: A Simulated-and-Real Data Augmentation for Autonomous Driving Perception under Challenging Weather. W7160456651
  • Sukwoo Jung et al. (2026). Toward Realistic Autonomous Driving Dataset Augmentation: A Real–Virtual Fusion Approach with Inconsistency Mitigation. 10.3390/s26030987
  • Martin Hahner et al. (2021). Fog Simulation on Real LiDAR Point Clouds for 3D Object Detection in Adverse Weather. 10.3929/ethz-b-000517119

weather / adverse / conditions

Measured impact of rain, fog and illumination on automotive sensors: multimodal adverse-weather datasets (MuFoRa), recognition technologies, sensor evaluation and performance monitoring, the PAVIN fog/rain platform and its digital twin, physics-based sensor simulation, road-spray modelling for LiDAR and field-tested all-weather shuttles.

• 使う op: koschmieder(視程と消散係数(Koschmieder)、any × any × scalar × any → any)、beta_from_mor(気象光学距離 → 消散係数、scalar → scalar)、fog_beta_from_profile(画像の明るさの縦断から霧の濃さ、any × any × scalar × scalar × scalar → table)、veiling_luminance(まぶしさの光幕輝度、any × any → any)、sight_stop_speed(見える距離で止まれる速さ、scalar × scalar × scalar → scalar)、env_params(天候の母数、入力なし → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Valentino Behret et al. (2026). MuFoRa: A Multimodal Dataset and the Impact of Adverse Weather on Camera and LiDAR. 10.1007/s42979-026-04976-9
  • Patrik Viktor et al. (2026). Sensors in Self-Driving Vehicles: A Detailed Literature Review and New Trends. 10.3390/s26072153
  • Yongsheng Qiu et al. (2024). Visual Perception Challenges in Adverse Weather for Autonomous Vehicles: A Review of Rain and Fog Impacts. 10.1109/itnec60942.2024.10733168
  • Neel Pratik Bhatt et al. (2023). WATonoBus: Field-Tested All-Weather Autonomous Shuttle Technology. 10.48550/arxiv.2312.00938
  • Charlotte Segonne et al. (2023). Qualification of the PAVIN Fog and Rain Platform and Its Digital Twin for the Evaluation of a Pedestrian Detector in Fog. 10.3390/jimaging9100211
  • Keisuke Yoneda et al. (2019). Automated driving recognition technologies for adverse weather conditions. 10.1016/j.iatssr.2019.11.005

weather / adverse / detection

Weather datasets, augmentation and robust detection: rain impact on cameras, stacked multi-weather city data, the Adver-City collaborative perception dataset, AirSim-W, data merging for YOLOv8, real-time photorealism enhancement for game engines (HyPER-GAN), foundation-model-guided detectors and GAN-based path planning in bad weather.

• 使う op: koschmieder(視程と消散係数(Koschmieder)、any × any × scalar × any → any)、beta_from_mor(気象光学距離 → 消散係数、scalar → scalar)、fog_beta_from_profile(画像の明るさの縦断から霧の濃さ、any × any × scalar × scalar × scalar → table)、veiling_luminance(まぶしさの光幕輝度、any × any → any)、sight_stop_speed(見える距離で止まれる速さ、scalar × scalar × scalar → scalar)、env_params(天候の母数、入力なし → table)、ほか 9 個: psnr、ssim、ms_ssim、ssim_map、fsim、gmsd、vifp、tid2013_evaluate、world_camera
• 代表論文(新しい順 + いちばん古い 1 本):
  • Sepideh Gohari et al. (2026). Real-World Perception for Autonomous Driving in Adverse Weather: Enhancing Standard Detectors via Foundation-Guided Auto-Annotation. 10.48550/arxiv.2609.25515
  • Danning Zhu et al. (2026). Object Detection for Autonomous Driving in Chinese Rural Scenes: An Experimental Study on Real-Synthetic Data Mixing and Model Evaluation. 10.48550/arxiv.2607.27058
  • Saibabu Merakanapalli et al. (2026). Autonomous Vehicle Safety in Adverse Weather and Emergency Conditions. 10.1109/ictmim68190.2026.11507456
  • Stefanos Pasios et al. (2026). HyPER-GAN: Hybrid Patch-Based Image-to-Image Translation for Real-Time Photorealism Enhancement in Game Engines. 10.48550/arxiv.2603.10604
  • Gerard de Mas Giménez et al. (2025). Multimodal sensing prototype for robust autonomous driving under adverse weather conditions. 10.1117/12.3062388
  • Elizabeth Bondi et al. (2018). AirSim-W. 10.1145/3209811.3209880

lidar / point / fog

LiDAR sensor modelling and point clouds under weather: empirical weather models for solid-state LiDAR, Unreal-based snowfall point-cloud generation, end-to-end LiDAR sensor modelling, denoising of rain/snow points, sim-to-real adaptation for segmentation, multi-scattering in fog, ToF LiDAR in fog chambers and physics-based augmentation (LISA).

• 使う op: koschmieder(視程と消散係数(Koschmieder)、any × any × scalar × any → any)、beta_from_mor(気象光学距離 → 消散係数、scalar → scalar)、fog_beta_from_profile(画像の明るさの縦断から霧の濃さ、any × any × scalar × scalar × scalar → table)、veiling_luminance(まぶしさの光幕輝度、any × any → any)、sight_stop_speed(見える距離で止まれる速さ、scalar × scalar × scalar → scalar)、env_params(天候の母数、入力なし → table)、ほか 7 個: lidar_spec、lidar_scan、ray_plane_range、ray_box_ranges、project_spherical、tof_multipath_phase、tof_transport_depth
• 代表論文(新しい順 + いちばん古い 1 本):
  • Ahmet Soyyigit et al. (2026). On Exploring Input Resolution Scaling For Anytime LiDAR Object Detection. 10.48550/arxiv.2607.08391
  • Himanshu Gupta et al. (2025). Evaluating LiDAR Perception Algorithms for All-Weather Autonomy. 10.3390/s25247436
  • Li et al. (2025). Investigating automotive sensor perception in rain. 10.82417/fhhs-x465
  • Xiaofang Luo et al. (2025). Train Snowfall LiDAR Point Cloud Generation Method Based on Unreal Engine and Physics Simulation. 10.1109/icirt66379.2025.11216782
  • Hyeong-Geun Kim (2025). SCOPE: Spatial Context-Aware Pointcloud Encoder for Denoising Under the Adverse Weather Conditions. 10.3390/app151810113
  • Zoltán Pusztai et al. (2017). Accurate Calibration of LiDAR-Camera Systems Using Ordinary Boxes. 10.1109/iccvw.2017.53

digital / twin / industry

57 本

Digital twins for autonomous-driving development and testing: twin-based test platforms (CARLA/Autoware vehicle-in-the-loop, PanoSim, mines and proving grounds), urban twin generation, sim-to-real transfer studies, multimodal datasets with full twin integration, and general digital-twin theory from manufacturing and smart cities.

• Model vs. twin: A digital twin is distinguished from a plain model by a live, bidirectional data link and synchronised state with its physical counterpart.
• Vehicle-in-the-loop twins: A real vehicle drives while virtual traffic and sensors are injected, combining real dynamics with safe scenarios.
• Twin generation: Open tools build road networks and 3D assets from OpenStreetMap, LiDAR and survey data for simulation.
• Virtual vs. physical transferability: Studies compare failures found in simulation with those on small-scale physical cars to quantify the reality gap.
• Digital siblings: Running tests in two simulators and combining verdicts reduces simulator-specific false failures.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• 自前の世界(真値つきの描画とセンサー): world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、gs_from_world(世界 → 3DGS、table → table)、carla_scene_world(CARLA の場面を自前の世界に(同じ場面を両方で)、table → table)、carla_depth_decode(CARLA の深度画像(RGB 符号)を m に、rgb → image2d)、carla_transform_matrix(CARLA の座標系(左手・度)↔ 自前の行列、any → matrix)、cr_read(CommonRoad のシナリオ XML を読む、any → table)、cr_drive(CommonRoad のシナリオを運転して採点、table → table)、osm_road_graph(OSM の道路網から地図、table → table)

twin / digital / transportation

Digital-twin systems for transportation testing: multi-layer transportation twins, pedestrian and connected-vehicle in-the-loop co-simulation, twin-based testing and data collection in extreme traffic, last-mile vehicle validation, reproducible longitudinal-control evaluation under complex weather and continuous testing with DTTF-Sim.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Yang Xu et al. (2026). A Reproducible Evaluation Method for Intelligent-Driving Longitudinal Control Under Complex Weather Through Operational Design Domain Parameter Perturbation. 10.3390/machines14040454
  • Zhigang Liang et al. (2025). DTTF-Sim: A Digital Twin-Based Simulation System for Continuous Autonomous Driving Testing. 10.3390/s25113447
  • Jiahe Cao et al. (2025). Poster: Multi-Layer Digital Transportation Twin. 10.1109/most65065.2025.00038
  • Zijin Wang et al. (2023). Towards Next Generation of Pedestrian and Connected Vehicle In-the-Loop Research: A Digital Twin Co-Simulation Framework. 10.1109/tiv.2023.3250353
  • Yueran Ding et al. (2022). A Digital Twin-based Testing and Data Collection System for Autonomous Driving in Extreme Traffic Scenarios. 10.1145/3579109.3579127
  • Lennart Luttkus et al. (2020). Scoomatic: Simulation and Validation of a Semi-Autonomous Individual Last-Mile Vehicle. 10.17185/duepublico/71204

digital / twin / virtual

Digital-twin platforms for AV development: digital siblings across simulators, AILiveSim, industry twin-driven design automation, VR human-vehicle interaction, campus and urban model building (OpenTwinMap), open vehicle-in-the-loop with CARLA and Autoware, mine and off-road proving-ground twins, low-fidelity twins and proactive safety validation (ADDT).

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Gábor Kusper et al. (2026). Digital-Twin-Oriented Virtual Training Environment for Agricultural Robot Navigation: A Vineyard Rover Case Study. 10.3390/s26123766
  • Lalitha Dabbiru et al. (2026). Digital twin of proving ground for off-road autonomous driving. 10.1117/12.3094700
  • Ari Pikkarainen et al. (2026). The Concept of a Digital Twin in the Arctic Environment. 10.3390/electronics15051001
  • Hwan-Hong Lee et al. (2026). Design and Validation of a Digital Twin Test Platform Leveraging Open-Source Autonomous Driving Systems. 10.1109/ojits.2026.3731513
  • Soyhan Yazgan (2026). Building Automotive Human–Machine Interfaces with Unreal Engine: Workflows, Toolchains, and Embedded Integration. W7162063276
  • Jerome Leudet et al. (2019). AILiveSim: An Extensible Virtual Environment for Training Autonomous Vehicles. 10.1109/compsac.2019.00074

sim2real / digital / twin

Sim-to-real transfer through digital twins: RL policies trained in simulation and moved to real vehicles via a twin, a survey of sim2real, digital twins and parallel intelligence, and an empirical comparison of virtual and physical-world testing of self-driving systems.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Xuemin Hu et al. (2023). How Simulation Helps Autonomous Driving: A Survey of Sim2real, Digital Twins, and Parallel Intelligence. 10.1109/tiv.2023.3312777
  • Kevin L. Voogd et al. (2023). Reinforcement Learning from Simulation to Real World Autonomous Driving using Digital Twin. 10.1016/j.ifacol.2023.10.1846
  • Andrea Stocco et al. (2022). Mind the Gap! A Study on the Transferability of Virtual Versus Physical-World Testing of Autonomous Driving Systems. 10.1109/tse.2022.3202311

digital / computing / edge

Two papers on digital twins in connected autonomous driving: DRL-based collaborative context offloading over C-V2X using a twin of vehicular contexts, and a full implementation and demonstration of twins for autonomous driving over wireless networks.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Kui Wang et al. (2023). Digital Twins for Autonomous Driving: A Comprehensive Implementation and Demonstration. 10.48550/arxiv.2401.08653
  • Kangkang Sun et al. (2023). Leveraging Digital Twin and DRL for Collaborative Context Offloading in C-V2X Autonomous Driving. 10.1109/tvt.2023.3333243

twin / digital / track

Datasets and low-cost platforms with digital-twin integration: the DrivIng multimodal dataset with a full twin of its route, and an open miniature Ackermann platform for end-to-end driving that pairs a physical car with simulation.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Gustavo Cláudio Karl Couto et al. (2026). A Low-Cost, Open Platform for End-to-End Autonomous Driving on a Miniature Ackermann Vehicle. 10.48550/arxiv.2609.04147
  • Dominik Rößle et al. (2026). DrivIng: A Large-Scale Multimodal Driving Dataset with Full Digital Twin Integration. 10.48550/arxiv.2601.15260
  • Dominik Rößle et al. (2026). DrivIng: A Large-Scale Multimodal Driving Dataset with Full Digital Twin Integration. W7125459488

digital / twin / twins

Digital-twin concepts and enablers across domains: realistic sensor simulation for twins, human and city twins, construction information systems, enabling technologies and challenges, multi-scale AV validation, twin-based vehicle testing theory, EV/AV twin reviews and edge computing.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Tanmay Vilas Samak et al. (2024). Towards Validation of Autonomous Vehicles Across Scales using an Integrated Digital Twin Framework. 10.1109/aim55361.2024.10637205
  • Barnaba Ubezio et al. (2023). Realistic sensor simulations for the digital twin. 10.1007/s00502-023-01156-y
  • Wasim A. Ali et al. (2023). A Review of Digital Twin Technology for Electric and Autonomous Vehicles. 10.3390/app13105871
  • Kacper Kubiak et al. (2022). Possible Applications of Edge Computing in the Manufacturing Industry—Systematic Literature Review. 10.3390/s22072445
  • Zhangjie Meng et al. (2022). The Vehicle Testing Based on Digital Twins Theory for Autonomous Vehicles. 10.1109/jrfid.2022.3211565
  • Aidan Fuller et al. (2020). Digital Twin: Enabling Technologies, Challenges and Open Research. 10.1109/access.2020.2998358

digital / twin / manufacturing

Digital-twin testing of autonomous driving plus general twin reviews from manufacturing: STPA-integrated defect detection, metamorphic testing with generative models, twin-based test systems and surveys, PanoSim twin testing, propulsion-drive twins and USD-based twin pipelines.

• 使う op: world_build(コースから 3-D の世界、table → table)、world_camera(色・ラベル・深度・面番号を一度に、table × matrix × matrix → table)、world_move(物体を動かす、table → any)、lidar_scan(同じ世界の LiDAR、mesh × table × matrix → table)、env_render(太陽・霧・雨・夜、table × matrix × matrix → table)、tile_stream(終わらない地図、table × table → table)、ほか 7 個: gs_from_world、carla_scene_world、carla_depth_decode、carla_transform_matrix、cr_read、cr_drive、osm_road_graph
• 代表論文(新しい順 + いちばん古い 1 本):
  • Dekun Zhang et al. (2026). A digital twin and STPA integrated framework for defect detection in autonomous driving systems. 10.1049/icp.2026.0521
  • Seok-Hyun Ahn (2025). A STUDY ON INTEGRATING DIGITAL TWIN PIPELINES USING USD & AI SIMULATION. 10.61413/qmbh8242
  • Tony Zhang et al. (2025). A Digital Twin Framework for Metamorphic Testing of Autonomous Driving Systems Using Generative Model. 10.1109/gcwkshps68340.2025.11591078
  • Duan Jianyu et al. (2023). Digital Twin Test Method for Autonomous Vehicles Based on PanoSim. 10.4271/2023-01-7056
  • Shenglin Wang et al. (2023). A deep learning-enhanced Digital Twin framework for improving safety and reliability in human–robot collaborative manufacturing. 10.1016/j.rcim.2023.102608
  • Anton Rassõlkin et al. (2019). Digital twin for propulsion drive of autonomous electric vehicle. 10.1109/rtucon48111.2019.8982326

planning / driving / learning

217 本

Motion planning and driving policies evaluated in simulation: closed-loop planning benchmarks (nuPlan, Bench2Drive, Waymax, Waymo Sim Agents), RL and imitation learning with sim-to-real transfer, end-to-end and diffusion-based trajectory planners, and learned or adversarial generation of safety-critical scenarios for planner testing.

• Open-loop vs. closed-loop: Displacement errors on logs predict closed-loop driving poorly; reactive closed-loop scores are the reference.
• Planning benchmark metrics: nuPlan-style scores combine collision, drivable-area, progress, comfort and TTC terms; CARLA uses driving score and infraction rates.
• Reactive agents: Rule-based (IDM) versus learned sim agents change benchmark rankings; realism is scored by distribution-matching metrics.
• Adversarial scenario generation: RL editing, kinematic gradients, guided diffusion, LLM agents and crash-grounded methods create collisions that stay plausible.
• Sim-to-real RL: Domain randomisation, modular abstraction and data-driven simulation transfer policies to real vehicles.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• 計画と制御(ルールベースの側): curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 学習・生成の運転モデル(採点する側は上の門) —— op なし(Fullseye の外)
• シナリオ試験と危険度の指標: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、turn_maneuver_check(右左折の採点、table → table)、skill_test_score(技能試験の減点、table → table)
• 交通流と他車の振る舞い: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、social_force_step(歩行者の Social Force、points × points × points → any)

scenarios / reinforcement / traffic

Learned traffic and safety-critical scenario generation for training and testing planners: RL-based scenario editing, multimodal adversarial generation, MetaDrive procedural composition, TrafficBots and TrafficGen learned agents, accident-prone generation via a learned traffic prior, KING kinematic gradients, AdvDiffuser and closed-loop corner-case platforms.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 15 個: turn_maneuver_check、skill_test_score、idm_platoon_simulate、idm_equilibrium_gap、driver_style、lateral_wobble、lane_change_permitted、passing_decision、social_force_step、curvature_speed_plan、pure_pursuit_curvature、stanley_steer、stop_line_plan、long_simulate、bicycle_model_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • Jiaheng Geng et al. (2025). Driving in Corner Case: A Real-World Adversarial Closed-Loop Evaluation Platform for End-to-End Autonomous Driving. 10.48550/arxiv.2512.16055
  • Jingyu Song et al. (2025). DriveCritic: Towards Context-Aware, Human-Aligned Evaluation for Autonomous Driving with Vision-Language Models. 10.48550/arxiv.2510.13108
  • Jiahao Wang et al. (2025). Drive&Gen: Co-Evaluating End-to-End Driving and Video Generation Models. 10.48550/arxiv.2510.06209
  • Yuting Xie et al. (2024). AdvDiffuser: Generating Adversarial Safety-Critical Driving Scenarios via Guided Diffusion. 10.48550/arxiv.2410.08453
  • Haolan Liu et al. (2024). Safety-Critical Scenario Generation Via Reinforcement Learning Based Editing. 10.1109/icra57147.2024.10611555
  • Wenhao Ding et al. (2020). Multimodal Safety-Critical Scenarios Generation for Decision-Making Algorithms Evaluation. 10.48550/arxiv.2009.08311

policy / end / learning

Policy learning and evaluation for end-to-end driving: world-model-based exploration, model-based policy adaptation, autoregressive token planners, learned trajectory scoring, counterfactual diagnosis of E2E and VLA policies under domain shift, plus an RL motion-cueing algorithm for driving simulators.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Ruolin Yang et al. (2026). Beyond the Leaderboard: Counterfactual Diagnosis of End-to-End and VLA Driving Policies Under Domain Shift. 10.48550/arxiv.2609.22582
  • Yaguang Li et al. (2026). Designing Versatile Samples for Learned Trajectory Scoring. 10.48550/arxiv.2609.01799
  • B. X. Ye et al. (2025). DAP: A Discrete-token Autoregressive Planner for Autonomous Driving. 10.48550/arxiv.2511.13306
  • Haojie Xin et al. (2025). Causal Contrastive Learning with Data Augmentations for Imitation-Based Planning. 10.1109/icra55743.2025.11128741
  • Xiaowei Huang et al. (2025). An Analytic Policy Gradient-Based Deep Reinforcement Learning Motion Cueing Algorithm for Driving Simulators. 10.1109/access.2025.3564597
  • Ce Hao et al. (2023). Improving Vehicle Trajectory Prediction with Online Learning. 10.1109/iv55152.2023.10186557

reinforcement / policy / learning

Reinforcement learning for driving and its transfer to reality: sim-to-real RL for full-size cars, bicycles and sailing boats, offline RL with latent skills, adversarial RL for safety-critical scenarios, data-driven simulation for robust control, multi-agent transfer, perception-imitating simulators and system-level safety design.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Shouma Amano et al. (2026). Sim-to-Real Aware End-to-End Learning Environment for Micromobility. 10.48550/arxiv.2609.28969
  • Thomas Steinecker et al. (2026). MILER: Semantic Mid-Level Representation for Sim-to-Real Reinforcement Learning in Unstructured Autonomous Driving. 10.48550/arxiv.2609.20747
  • Xincong Hu et al. (2026). Threat-guided Policy-aware Scene Perturbation for Safe Autonomous Driving with Online Reinforcement Learning. 10.48550/arxiv.2608.10403
  • Kiki Johanna Alberdina Bink et al. (2026). Autonomous sailing with sim-to-real reinforcement learning. 10.1016/j.engappai.2026.115804
  • Luke Rowe et al. (2026). Scaling Self-Play for End-to-End Driving. 10.48550/arxiv.2606.19641
  • Xinlei Pan et al. (2017). Virtual to Real Reinforcement Learning for Autonomous Driving. 10.48550/arxiv.1704.03952

end / trajectory / planning

End-to-end trajectory planning from sensors: trajectory selection and refinement (DriveSuprim, DiffRefiner, ResAD), coupled perception-in-planning, BEV transformers with Bezier optimisation, trajectory-guided control, MLLM reasoning planners, world-model plus VLM frameworks, personalised and safety-reasoning E2E driving and RL fine-tuning of VLA models.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Sungmo Ku et al. (2026). A Multi-Task Framework for Vehicle Trajectory and Disturbance Classification in Autonomous Driving Scenarios. 10.3390/ijgi15100443
  • Yuan Gao et al. (2026). PlannerForge: LLM Agents for Scenario-Based Testing of Motion Planners in Autonomous Driving. 10.48550/arxiv.2609.08965
  • Yikai Wu (2026). From Proxy Learning to Driving Decisions: A Transfer-Based Framework for Evaluating Future-Aware Autonomous Driving Planners. 10.48550/arxiv.2609.02688
  • Xinlin Wang et al. (2026). WA-JEPA: Rethinking the Video JEPA Paradigm for World-Action Modeling in Autonomous Driving. 10.48550/arxiv.2608.20974
  • Simón Patiño Idarraga et al. (2026). Herding End-to-End Autonomous Driving via Neuro-Symbolic Safety Guards. 10.48550/arxiv.2608.11451
  • Liu Yifei et al. (2021). Review of end-to-end motion planning for autonomous driving with visual perception. 10.11834/jig.200276

collision / critical / scenario

Generation of collision and safety-critical scenarios for AV testing: LLM-driven ChatScene, retrieval-augmented online collision seeking, Any2Critical from arbitrary real contexts, benchmarking adversarial agent strategies, crash-grounded generation (RCG), map-agnostic tree search and time-reversed generation from collision snapshots.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Taehyung Kim et al. (2026). Collision Snapshot Guided Time-Reversed Safety-Critical Scenario Generation. 10.48550/arxiv.2609.06433
  • Carmen Mei-Ling Frischknecht-Gruber et al. (2026). BAAS: Benchmarking Adversarial Agent Strategies : a comparative study of gradient-based, reinforcement, and evolutionary paradigms for safety-critical scenario generation. 10.21256/zhaw-37049
  • Yao Huang et al. (2026). Any2Critical: Safety-Critical Scenario Generation from Arbitrary Real-World Driving Contexts. 10.1609/aaai.v40i42.40861
  • Wenyun Li et al. (2026). Map-Agnostic And Interactive Safety-Critical Scenario Generation via Multi-Objective Tree Search. 10.48550/arxiv.2603.03978
  • Association for Artificial Intelligence 2026 et al. (2026). Any2Critical: Safety-Critical Scenario Generation from Arbitrary Real-World Driving Contexts. 10.48448/27r7-em03
  • Jiawei Zhang et al. (2024). ChatScene: Knowledge-Enabled Safety-Critical Scenario Generation for Autonomous Vehicles. 10.48550/arxiv.2405.14062

merging / motion / micro

Small leaf on merging and safe motion planning: predicted-occupancy online safety verification, the Bench4Merge benchmark for dense-traffic merging with micro-interactive vehicles, state-lattice planning and safety-critical scenario generation with vision-based driving.

• 使う op: idm_platoon_simulate(IDM の車列、any → table)、idm_equilibrium_gap(IDM の定常車間(閉形式)、any → any)、driver_style(運転の癖を個体ごとに、入力なし → table)、lateral_wobble(車線内のふらつき、入力なし → signal)、lane_change_permitted(進路変更の可否、table → table)、passing_decision(すれ違いの判断、入力なし → any)、ほか 7 個: social_force_step、curvature_speed_plan、pure_pursuit_curvature、stanley_steer、stop_line_plan、long_simulate、bicycle_model_step
• 代表論文(新しい順 + いちばん古い 1 本):
  • Chengyuan Zheng et al. (2026). Predicted occupancy map-based online safety verification and resilient motion planning for autonomous driving. 10.1016/j.trc.2026.105577
  • Zhengming Wang et al. (2025). Bench4Merge: A Comprehensive Benchmark for Merging in Realistic Dense Traffic with Micro-Interactive Vehicles. 10.1109/iros60139.2025.11247075
  • Zhengming Wang et al. (2024). Bench4Merge: A Comprehensive Benchmark for Merging in Realistic Dense Traffic with Micro-Interactive Vehicles. 10.48550/arxiv.2410.15912
  • Wang et al. (2024). Enhancing safety and generalization for autonomous driving : safety-critical scenario generation and vision-based resolution. W7113295532
  • Shuiying Wang (2015). State Lattice-based Motion Planning for Autonomous On-Road Driving. 10.17169/refubium-7243

planning / loop / closed

The largest planning leaf: learning-based planners and the benchmarks and simulators that evaluate them.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Victoria Smirnova et al. (2026). TrafficSignBench: Rule-Centric Closed-Loop Evaluation of Traffic-Sign Compliance in Autonomous Driving. 10.48550/arxiv.2609.38463
  • Ziyang Leng et al. (2026). DreamStream: Towards Policy-Oriented Generative Simulation for End-to-End Driving. 10.48550/arxiv.2609.26792
  • Arka Pal et al. (2026). One Diffusion Model, Two Roles: Guided Trajectory Planning and Safety-Critical Scenario Generation in Closed-Loop Simulation. 10.48550/arxiv.2609.04921
  • Ziang Wei et al. (2026). When Shared Rollouts Fail in Defensive Driving Evaluation: A NAVSIM Score Basis Audit. 10.48550/arxiv.2608.04896
  • Jinyu Miao et al. (2026). nuTruck: Benchmarking Autonomous Driving Planning for Distributed Electric-drive Trucks. 10.48550/arxiv.2607.13704
  • Nicholas Rhinehart et al. (2021). Contingencies from Observations: Tractable Contingency Planning with Learned Behavior Models. 10.1109/icra48506.2021.9561683

expert / imitation / policy

Imitation learning and expert design for end-to-end urban driving: RL coaches as experts (Roach), general reinforced imitation, human-in-the-loop RL, traffic-rule compliance, learner-expert asymmetry, the imitation-gap benchmark IGDrivSim, the PCLA testing framework for CARLA agents, mixture-of-experts planners and reliable world simulation.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Long Nguyen et al. (2025). LEAD: Minimizing Learner-Expert Asymmetry in End-to-End Driving. 10.48550/arxiv.2512.20563
  • Wan et al. (2025). GEMINUS: Dual-aware Global and Scene-Adaptive Mixture-of-Experts for End-to-End Autonomous Driving. 10.48550/arxiv.2507.14456
  • Masoud Jamshidiyan Tehrani et al. (2025). PCLA: A Framework for Testing Autonomous Agents in the CARLA Simulator. 10.1145/3696630.3728577
  • Masoud Jamshidiyan Tehrani et al. (2025). PCLA: A Framework for Testing Autonomous Agents in the CARLA Simulator. 10.48550/arxiv.2503.09385
  • Jia-Zhi Yang et al. (2025). ReSim: Reliable World Simulation for Autonomous Driving. 10.52202/085713-5591
  • Songan Zhang (2021). Synthesis and Evaluation of Automated Vehicles. 10.7302/1485

reconstruction / scene / view

132 本

Neural scene reconstruction and novel view synthesis for sensor simulation: NeRF variants for street scenes, LiDAR and radar fields, 3D Gaussian Splatting for static and dynamic urban scenes, diffusion-assisted view synthesis, asset insertion and editing, and closed-loop simulators built on reconstructed worlds (HUGSIM, S-NeRF++, NeuRAD).

• View-synthesis quality metrics: PSNR, SSIM and LPIPS on held-out frames, plus FID/FVD for generated video; report on interpolated and extrapolated (lane-shifted) views separately.
• Geometry metrics: Depth error against LiDAR (abs-rel, RMSE) and Chamfer distance check that reconstructions are metrically usable.
• Dynamic decomposition: Scenes are split into static background and per-actor models, using boxes, motion masks or self-supervised flow, so actors can be moved.
• Multi-sensor rendering: LiDAR and radar neural fields, rolling shutter and multi-camera models are required for full sensor simulation.
• Extrapolation weakness: Quality drops sharply away from the recorded trajectory; diffusion priors and map or depth priors mitigate it.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• ニューラル再構成と 3DGS: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)

nerf / radiance / neural

Neural radiance fields and related neural rendering for driving scenes and beyond: urban radiance fields, NeuRAD for multi-sensor rendering, LiDAR-aided street NeRFs, radar neural fields, in-the-wild object reconstruction, point-based street rendering, NeRF-LiDAR point generation, editable scene simulation with LLM agents and NeRF uncertainty visualisation.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Yuzhou Zhou (2026). Unifying Street Scene Point Cloud Semantic Segmentation with Deformable Mesh-based Neural Representation. 10.5194/isprs-archives-xlix-b2-2026-327-2026
  • Zheng Zeng (2026). Toward Generative Rendering: Controllable Image Synthesis with High Photorealism and Low Cost. W7213486193
  • Noah Lewis et al. (2025). NePO: Neural Point Octrees for Large‐Scale Novel View Synthesis. 10.1111/cgf.70287
  • Kirsten W. H. Maas et al. (2025). NeRVis: Neural Radiance Field Model-Uncertainty Visualization. 10.1016/j.cag.2025.104511
  • Yun Liao et al. (2025). A Survey on Neural Radiance Fields. 10.1145/3758085
  • Ka Wai Tsoi (2016). Improve OpenMVG and create a novel algorithm for novel view synthesis from point clouds. W2611242618

synthesis / view / gaussian

Street-view synthesis with Gaussian splatting and diffusion: self-supervised dynamic street reconstruction, explicit motion modelling, controllable video diffusion (StreetCrafter), diffusion priors for sparse views (SGD, MuDG), semantic and temporal constraints, LiDAR-reflectance guidance, neural radar fields, asset insertion and geometry-informed world models.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Mochuan Zhan et al. (2026). Metric-consistent road-surface novel view synthesis under road-centric sparse-view acquisition. 10.1016/j.cag.2026.104732
  • Hongbo Lu et al. (2026). VisionNVS: Self-Supervised Inpainting for Novel View Synthesis under the Virtual-Shift Paradigm. 10.48550/arxiv.2603.17382
  • Ziyu Chen et al. (2026). LR-SGS: Robust LiDAR-Reflectance-Guided Salient Gaussian Splatting for Self-Driving Scene Reconstruction. 10.48550/arxiv.2603.12647
  • Qingwu Duan et al. (2026). Dynamic Street-Scene Reconstruction with Semantic Priors and Temporal Constraints. 10.3390/app16052421
  • Soujin Cho et al. (2026). Accuracy of three-dimensional Gaussian Splatting for virtual crime scene reconstruction. 10.3389/fcomp.2026.1755361
  • Yujiao Shi et al. (2021). Self-Supervised Visibility Learning for Novel View Synthesis. 10.48550/arxiv.2103.15407

rendering / occlusions / gaussian

A single paper (in three copies) on multimodal Gaussian splatting that fuses camera images with other sensing such as radar or LiDAR for robust 3D scene rendering where RGB alone is unreliable.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Chi-Shiang Gau et al. (2026). 3D Scene Rendering with Multimodal Gaussian Splatting. 10.48550/arxiv.2602.17124
  • Chi-Shiang Gau et al. (2026). 3D Scene Rendering with Multimodal Gaussian Splatting. W7130762625
  • Chi-Shiang Gau et al. (2025). 3D Scene Rendering with Multimodal Gaussian Splatting. 10.1109/ieeeconf67917.2025.11443659

view / gaussian / splatting

Large-scale and sparse-view reconstruction: sparse-view 3DGS reviews, fine surface reconstruction in large scenes, planar street-surface splatting, crowd-sourced NeRF from production vehicles, hierarchical dynamic Gaussians, thermal splatting, geo-localisation via synthesis, grid-guided and focal NeRFs and occupancy-guided unbounded scene generation.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Xiaoyu Ye et al. (2026). InfiniVerse: Occupancy Guided Unbounded Scene Generation for Autonomous Driving. 10.48550/arxiv.2606.31109
  • Fudong Ge et al. (2026). HDGS: Hierarchical Dynamic Gaussian Splatting for Urban Driving Scenes. 10.1609/aaai.v40i6.42419
  • Xiaokun Ding et al. (2025). Cross-View Geo-Localization via 3D Gaussian Splatting-Based Novel View Synthesis. 10.3390/rs17223673
  • Shihan Chen et al. (2025). 3D Gaussian Splatting for Fine-Detailed Surface Reconstruction in Large-Scale Scene. 10.1109/iros60139.2025.11247350
  • Shihan Chen et al. (2025). 3D Gaussian Splatting for Fine-Detailed Surface Reconstruction in Large-Scale Scene. 10.48550/arxiv.2506.17636
  • Xiang Guo et al. (2020). Novel View Synthesis from only a 6-DoF Camera Pose by Two-stage Networks. 10.48550/arxiv.2010.11468

survey / nerf / splatting

Surveys and reviews of NeRF and 3D Gaussian Splatting: general 3DGS surveys, driving-specific reconstruction reviews, NeRF in autonomous driving, real-world NeRF challenges, dynamic-scene 3DGS reviews and NeRF versus photogrammetry comparisons.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Wenhui Xiao et al. (2026). Neural Radiance Fields for the Real World: A Survey. 10.1145/3837758
  • Guikun Chen et al. (2026). A Survey on 3D Gaussian Splatting. 10.1145/3807511
  • Zhenhan Lian et al. (2025). A Review of Dynamic Scene Reconstruction Based on 3D Gaussian Splatting. 10.3724/sp.j.1089.2025-00251
  • Huixin Zhu et al. (2024). Scene reconstruction techniques for autonomous driving: a review of 3D Gaussian splatting. 10.1007/s10462-024-10955-4
  • Ben Fei et al. (2024). 3D Gaussian Splatting as a New Era: A Survey. 10.1109/tvcg.2024.3397828
  • Valeria Croce et al. (2024). Comparative Assessment of Neural Radiance Fields and Photogrammetry in Digital Heritage: Impact of Varying Image Conditions on 3D Reconstruction. 10.3390/rs16020301

views / radiance / fields

Novel view synthesis methods with priors for outdoor and driving views: motion-blur-robust NeRF, geometry-free scene representation transformers, map priors (MapNeRF), depth priors, super-resolution NeRF from sparse views, single-image synthesis, S-NeRF for streets, car radiance fields from in-the-wild images and multi-sensor conditioning.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Yoshiyuki Shirasaki et al. (2026). EDoF-NeRF: extended depth-of-field neural radiance fields using a coded aperture camera. 10.48550/arxiv.2606.18826
  • Zhengfei Kuang et al. (2026). Effective Multi-sensor Conditioning for Street-view Novel-view Synthesis. 10.48550/arxiv.2606.01590
  • Changbo Gao et al. (2025). MBS-NeRF: reconstruction of sharp neural radiance fields from motion-blurred sparse images. 10.1038/s41598-025-88614-z
  • Jun Wang et al. (2025). SRNeRF: Super-Resolution Neural Radiance Fields for Autonomous Driving Scenario Reconstruction from Sparse Views. 10.3390/wevj16020066
  • Chen Wang et al. (2023). Digging into Depth Priors for Outdoor Neural Radiance Fields. 10.1145/3581783.3612306
  • Yuxin Hou et al. (2021). Novel View Synthesis via Depth-guided Skip Connections. 10.48550/arxiv.2101.01619

scene / scenes / reconstruct

Sensor-data synthesis for dynamic driving scenes: SurfelGAN surfel reconstruction with GAN refinement, DrivingGaussian composite Gaussians, priority-adaptive and vision-centric Gaussian splatting, instance-aware 3DGS from dashcam video and 4D simulation with video generation (Stag-1).

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • An Ying et al. (2026). PAGS: Priority-Adaptive Gaussian Splatting for Dynamic Driving Scenes. 10.1109/icassp55912.2026.11460978
  • Hongyuan Liu et al. (2025). InstDrive: Instance-Aware 3D Gaussian Splatting for Driving Scenes. 10.48550/arxiv.2508.12015
  • Lening Wang et al. (2024). Stag-1: Towards Realistic 4D Driving Simulation with Video Generation Model. 10.48550/arxiv.2412.05280
  • Jiusi Li et al. (2024). VC-Gaussian: Vision-Centric Gaussian Splatting for Dynamic Autonomous Driving Scenes. 10.1109/itsc58415.2024.10919531
  • Xiaoyu Zhou et al. (2024). DrivingGaussian: Composite Gaussian Splatting for Surrounding Dynamic Autonomous Driving Scenes. 10.1109/cvpr52733.2024.02044
  • Zhenpei Yang et al. (2020). SurfelGAN: Synthesizing Realistic Sensor Data for Autonomous Driving. 10.48550/arxiv.2005.03844

gaussian / reconstruction / splatting

Gaussian-splatting-based driving scene reconstruction and simulators: synthetic scene editing evaluation, multi-camera-model reconstruction, trajectory-flexible rendering, Bezier-curve dynamic Gaussians, constrained AutoSplat, DENSER, V2X cooperative reconstruction, LiDAR-supervised highway scenes, feed-forward reconstruction, unified sensor simulation with rolling shutter and the HUGSIM closed-loop simulator.

• 使う op: gs_from_world(メッシュの世界 → ガウス、table → table)、gs_render(ガウスの描画、table × matrix → table)、gs_update(ガウスの更新、table × table → table)、gs_read_file(3DGS の読み込み、入力なし → table)、humanoid_impostors(向き × 位相の事前描画(安い代替)、table → table)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Nikolay Patakin et al. (2026). Unified Sensor Simulation for Autonomous Driving. 10.24963/ijcai.2026/168
  • NVIDIA et al. (2026). Instant NuRec: Feed-Forward 3D Gaussian Reconstruction for Driving Scene Simulation. 10.48550/arxiv.2607.14203
  • Ali Nouri et al. (2026). From Concept to Capability: Evaluating 3D Gaussian Splatting for Synthetic Scene Editing in Autonomous Driving. 10.1007/978-3-032-34867-8_11
  • Ali Nouri et al. (2026). From Concept to Capability: Evaluating 3D Gaussian Splatting for Synthetic Scene Editing in Autonomous Driving. 10.48550/arxiv.2605.01995
  • Yedong Shen et al. (2026). GA-GS: Generation-Assisted Gaussian Splatting for Static Scene Reconstruction. 10.48550/arxiv.2604.04331
  • Yun Chen et al. (2021). GeoSim: Realistic Video Simulation via Geometry-Aware Composition for Self-Driving. 10.48550/arxiv.2101.06543

driving / learning / autonomous

232 本

Learning-based driving and the simulators, world models and synthetic data that feed it: end-to-end sensor-fusion policies evaluated in CARLA, vision-language(-action) driving models, generative driving world models and video simulators, neural sensor simulators (LiDARsim, UniSim), corner-case detection and generation, and sim-to-real transfer across robots and vehicles.

• End-to-end closed-loop evaluation: CARLA Leaderboard and Bench2Drive driving score, route completion and infractions per km are the main measures for E2E and VLA policies.
• World models as simulators: Action-conditioned video or occupancy generators predict futures; judge them by FVD/FID, controllability and physical consistency.
• Neural sensor simulation: LiDARsim and UniSim rebuild real logs into assets so new sensor data can be rendered for closed-loop tests.
• Corner-case taxonomy: Corner cases are classified from sensor and content level up to scenario level, guiding detection and generation.
• Curse of rarity: Safety-critical events are rare in high-dimensional spaces, so naive data collection and training under-represent them.

使う op(テーマ → op の対応表から。ノートで型と例を確かめてから使う):

• 学習・生成の運転モデル(採点する側は上の門) —— op なし(Fullseye の外)
• シナリオ試験と危険度の指標: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、turn_maneuver_check(右左折の採点、table → table)、skill_test_score(技能試験の減点、table → table)
• sim-to-real の比較(同じ場面を対で測る): psnr(画素の差、image2d × image2d → scalar)、ssim(構造の類似、image2d × image2d → scalar)、ms_ssim(多段の構造類似、image2d × image2d → scalar)、ssim_map(どこが違うかの地図、image2d × image2d → image2d)、fsim(位相一致と勾配の知覚的類似(FSIM)、image2d × image2d → scalar)、gmsd(勾配の類似の偏差(GMSD)、image2d × image2d → scalar)、vifp(情報忠実度(VIFp)、image2d × image2d → scalar)、tid2013_evaluate(指標を人の評価(TID2013)と順位相関で採点、any × any → table)、world_camera(真値つきの対の描画、table × matrix × matrix → table)
• 計画と制御(ルールベースの側): curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)

navigation / unmanned / racing

A heterogeneous leaf on simulation for navigation and training beyond road cars: nano-drone racing and USV navigation with sim-to-real DRL, flapping-wing and unmanned-craft simulators, AI tutoring in driving simulators, indoor navigation, disaster reconnaissance, synthetic data in robotics, autonomous racing platforms and mesh fidelity requirements for lidar simulation.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Haichuan Li et al. (2026). A Simulation Platform for Flapping-Wing Vehicles. 10.48550/arxiv.2606.02370
  • Harishik Dev Singh Jamwal et al. (2026). Comprehensive Survey on Autonomous Disaster Reconnaissance: A Comparative Analysis of UAVs and UGVs. 10.3390/telecom7020031
  • Wenbin Gan et al. (2026). Intelligent Tutoring in a Driving Simulator: Enhancing Driving Proficiency With AI-Driven Skill Assessment and Personalized Coaching Generation. 10.1145/3742413.3789060
  • Tanay Choudhary (2025). The Role of Synthetic Data in Robotics: Accelerating Development and Innovation. 10.32628/cseit251112318
  • Christopher T. Goodin et al. (2024). Geometric Fidelity Requirements for Meshes in Automotive Lidar Simulation. 10.3390/virtualworlds3030014
  • Naser El‐Sheimy et al. (2021). Indoor navigation: state of the art and future trends. 10.1186/s43020-021-00041-3

video / diffusion / videos

Generative world models and data tools for driving: neural map priors, NeRF real-to-sim gap studies, 4D occupancy generation (OccSora), long-video diffusion, DiT-based controllable video generation, streaming scene generators, navigation world models, long-term driving world models, real-time dynamic LiDAR ray casting and the SimVeRi synthetic re-identification dataset from CARLA-SUMO co-simulation.

• 代表論文(新しい順 + いちばん古い 1 本):
  • Anonymous Anonymous (2026). Raw simulation traces for “How Close Was That? Grounding Continuous Test Oracles for Autonomous Vehicles in Counterfactual Harm”. 10.5281/zenodo.22736471
  • Anonymous Anonymous (2026). Raw simulation traces for “How Close Was That? Grounding Continuous Test Oracles for Autonomous Vehicles in Counterfactual Harm”. 10.5281/zenodo.22736472
  • Zengzhi-Zhang et al. (2026). SimVeRi code. 10.5281/zenodo.19207142
  • Zengzhi Zhang et al. (2026). SimVeRi code. 10.5281/zenodo.21980102
  • Zengzhi Zhang et al. (2026). SimVeRi: A synthetic multi-camera vehicle re-identification dataset with spatiotemporal metadata, controlled appearance ambiguity, and aerial–ground views. 10.5281/zenodo.21037242
  • Soumi Das et al. (2021). TMCOSS: Thresholded Multi-Criteria Online Subset Selection for Data-Efficient Autonomous Driving. 10.1109/iccv48922.2021.00628

action / learning / future

Vision-language-action models, world-action models and sim-to-real learning: driving world models unified with planning (DrivingGPT, DriveVLA-W0), VLA efficiency and grounding, memory-based driving, foundation models for AVs, interactive generative video, sensor-attack systematisation, HD map generation and sim-to-real transfer surveys for robot controllers.

• 代表論文(新しい順 + いちばん古い 1 本):
  • Christian Löwens (2026). Driving on Memory: checkpoints and memory banks. 10.5281/zenodo.22959057
  • Christian Löwens (2026). Driving on Memory: checkpoints and memory banks. 10.5281/zenodo.22959058
  • Erik Deinzer et al. (2026). PRIME: Perception Feedback with Situational Memory Embeddings in VLA Models. 10.48550/arxiv.2609.22040
  • Hui Zhou (2026). Advances in ROS-Based Robot Simulation Technology: A Mini Review. 10.62762/ticps.2026.304297
  • Dogun Kim et al. (2026). RAF-VLA: Representation Alignment with the Future for End-to-End Autonomous Driving. 10.48550/arxiv.2609.17728
  • Jacques Kaiser et al. (2019). Embodied Synaptic Plasticity With Online Reinforcement Learning. 10.3389/fnbot.2019.00081

corner / cases / case

Corner cases in autonomous driving and embodied AI: detection and generation overviews, visual analytics for interpretation, co-occurrence knowledge graphs, full-physics 77 GHz radar simulation for pedestrian detection, simulator-based survival analysis, Zipfian environments, the W-CODA workshop and arguments on physical AI and cognitive planning.

• 使う op: ttc_truth(衝突までの時間の真値、image2d × matrix × matrix → table)、ttc_from_range(距離の列からの TTC、入力なし → scalar)、rss_longitudinal_check(RSS の縦の安全距離、table → table)、rss_lateral_check(RSS の横の安全距離、table → table)、dilemma_zone(黄信号のジレンマゾーン(閉形式)、any → table)、occlusion_safe_speed(死角から止まれる速さ、any → any)、ほか 2 個: turn_maneuver_check、skill_test_score
• 代表論文(新しい順 + いちばん古い 1 本):
  • Yunji Liang et al. (2026). Corner Case Detection and Generation for Autonomous Driving: An Overview. 10.1109/tits.2026.3688488
  • Partha Pratim Ray (2025). Physical AI: Bridging the Sim-to-Real Divide Toward Embodied, Ethical, and Autonomous Intelligence. 10.36227/techrxiv.176238257.73914361/v1
  • S. Jam bar sang (2025). Constructing the Umwelt: Cognitive Planning through Belief-Intent Co-Evolution. 10.48550/arxiv.2511.05540
  • Sicheng Zhan et al. (2025). Dealing with corner cases in occupant-centric control: do physics-informed models help?. 10.26868/25222708.2025.1808
  • Kai Chen et al. (2025). ECCV 2024 W-CODA: 1st Workshop on Multimodal Perception and Comprehension of Corner Cases in Autonomous Driving. 10.48550/arxiv.2507.01735
  • Ushemadzoro Chipengo (2019). Full physics simulation of terrain‐adaptive 77 GHz automotive radar for early pedestrian detection. 10.1002/mop.31714

avatar / continual / poses

Small leaf mixing pedestrian corner-case augmentation via pose-distribution control, maintainable NeRF avatars with continual learning, informed priors for trajectory prediction and federated NeRF.

• 代表論文(新しい順 + いちばん古い 1 本):
  • Shengbo Gu et al. (2025). MaintaAvatar: A Maintainable Avatar Based on Neural Radiance Fields by Continual Learning. 10.1609/aaai.v39i3.32327
  • Gu et al. (2025). MaintaAvatar: A Maintainable Avatar Based on Neural Radiance Fields by Continual Learning. 10.48550/arxiv.2502.02372
  • Yintian Zhang et al. (2024). Federated Neural Radiance Field for Distributed Intelligence. 10.48550/arxiv.2406.10474
  • Yunhao Nie et al. (2023). Data Augmentation for Pedestrians in Corner Case: A Pose Distribution Control Approach. 10.1109/itsc57777.2023.10422639
  • Christian Schlauch et al. (2022). Informed Priors for Knowledge Integration in Trajectory Prediction. 10.48550/arxiv.2211.00348

language / driving / loop

Language and multimodal models in closed-loop driving: SimLingo language-action alignment, LLM4AD reviews, LanguageMPC, dual-system efficiency (ETA), RAG for safety, CoT-based VLA, compositional traffic-rule benchmarks, Bench2Drive-VL, illusion robustness benchmarks, collaborative perception, UniSim neural closed-loop sensor simulation and evasion attacks in CARLA.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Alexey Zakharov et al. (2026). Less Language, More Latents: Annotation-Efficient VLAs for Driving. 10.48550/arxiv.2609.27747
  • Tianyuan Zhang et al. (2026). Benchmarking the Robustness of Autonomous Driving to Environmental Illusions: A Lane Perception Perspective. 10.1109/tpami.2026.3710189
  • Hao Vo et al. (2026). OmniSpace: Efficient Geometry Awareness for Autonomous Vehicles MLLMs. 10.48550/arxiv.2606.22617
  • Jin Yao et al. (2026). VLGA: Vision-Language-Geometry-Action Models for Autonomous Driving. 10.48550/arxiv.2606.12396
  • Zhiyu Huang et al. (2026). nuReasoning: A Reasoning-Centric Dataset and Benchmark for Long-Tail Autonomous Driving. 10.48550/arxiv.2605.31572
  • Nayun Xu et al. (2018). Autonomous Driving in Reality with Reinforcement Learning and Image Translation. 10.48550/arxiv.1801.05299

end / sensor / autonomous

End-to-end driving with sensor fusion and the simulators used to train and test it: TransFuser, ReasonNet, interpretable and multimodal E2E policies, LiDARsim, domain-randomised obstacle avoidance, radar-model measurement challenges, robustness benchmarks (Bench2Drive-Robust), hidden dataset biases, and simulators for robotics, drones, racing and space sensing.

• 使う op: curvature_speed_plan(曲率から速度の計画、any → table)、pure_pursuit_curvature(pure pursuit、any → table)、stanley_steer(Stanley、any → table)、stop_line_plan(停止線までの計画、scalar × scalar → table)、long_simulate(縦の運動、scalar × scalar × any → table)、bicycle_model_step(自転車モデル、any → any)
• 代表論文(新しい順 + いちばん古い 1 本):
  • Dhieddine Barhoumi et al. (2026). Compressing Transformer-Based Sensor Fusion Models for Autonomous Driving Using Curriculum-Based Multi-Task Knowledge Distillation. 10.3390/app16189359
  • Shunsuke Ito et al. (2026). BA-TRACE: Boundary-Aware Trace Reconstruction for Scenario-Based Evaluation of Mixed AUTOSAR Adaptive and ROS 2 Vehicular Embedded Systems. 10.48550/arxiv.2609.19699
  • Davide Santolin et al. (2026). Accelerating stochastic processes through nonequilibrium driving: Thermodynamic constraints on the maximum speed-up. 10.48550/arxiv.2609.03179
  • Ke Guo et al. (2026). iPad: Iterative Proposal-Centric End-to-End Autonomous Driving. 10.1109/lra.2026.3723334
  • Chong He et al. (2026). DriveVer: Lightweight Trajectory Evaluator as Test-Time Verifier for Autonomous Driving. 10.48550/arxiv.2607.00399
  • Masaru Shimizu et al. (2015). A Realistic RoboCup Rescue Simulation Based on Gazebo. 10.1007/978-3-319-29339-4_27

synthetic / learning / intelligence

Synthetic data and corner cases at a conceptual level: the curse of rarity for AV safety, holistic anomaly detection, domain-randomised RL for surfacing corner cases (DR2L), the role and politics of synthetic data, small-data ML, open-source autonomy experience, Unreal Engine suitability and long-term AV reviews.

• 使う op: psnr(画素の差、image2d × image2d → scalar)、ssim(構造の類似、image2d × image2d → scalar)、ms_ssim(多段の構造類似、image2d × image2d → scalar)、ssim_map(どこが違うかの地図、image2d × image2d → image2d)、fsim(位相一致と勾配の知覚的類似(FSIM)、image2d × image2d → scalar)、gmsd(勾配の類似の偏差(GMSD)、image2d × image2d → scalar)、ほか 3 個: vifp、tid2013_evaluate、world_camera
• 代表論文(新しい順 + いちばん古い 1 本):
  • Styliani-Christina Fragkouli et al. (2026). An ELIXIR scoping review on domain-specific evaluation metrics for synthetic data in life sciences. 10.1093/nargab/lqag012
  • Andrei Vlad Man et al. (2026). RoD-TAL: A Benchmark for Answering Questions in Romanian Driving License Exams. 10.18653/v1/2026.findings-eacl.295
  • Jakub Július Šmýkal (2026). Platform for HIL testing of sensoric applications. W7165975317
  • Shivani Kapania et al. (2025). Examining the Expanding Role of Synthetic Data Throughout the AI Development Pipeline. 10.1145/3715275.3732005
  • Henry X. Liu et al. (2025). Autonomous Vehicles: A Critical Review (2004-2024) and a Vision for the Future. 10.36227/techrxiv.174857767.78237989/v1
  • Michalis Korakakis et al. (2018). A short survey on modern virtual environments that utilize AI and synthetic data. W2940784521

op → クラスタ(逆引き)

この文献層で「使う op」に挙がった op と、それが出たクラスタ(上位 › 葉)。op から「どの知識の文脈で使うか」を引く。

op	出たクラスタ
beta_from_mor	weather / detection / synthetic
bicycle_model_step	planning / driving / learning; driving / learning / autonomous
bus_stop_rate	sampling / importance / rare; scenario / testing / automated
carla_depth_decode	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
carla_scene_world	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
carla_transform_matrix	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
cr_drive	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
cr_read	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
curvature_speed_plan	planning / driving / learning; driving / learning / autonomous
dilemma_zone	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
doppler_shift	weather / detection / synthetic
doppler_track	weather / detection / synthetic
driver_style	traffic / simulation / vehicle; planning / driving / learning
env_params	weather / detection / synthetic
env_render	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
fmcw_range_profile	weather / detection / synthetic
fog_beta_from_profile	weather / detection / synthetic
fsim	weather / detection / synthetic; driving / learning / autonomous
gmsd	weather / detection / synthetic; driving / learning / autonomous
gs_from_world	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry; reconstruction / scene / view
gs_read_file	reconstruction / scene / view
gs_render	reconstruction / scene / view
gs_update	reconstruction / scene / view
humanoid_impostors	reconstruction / scene / view
idm_equilibrium_gap	traffic / simulation / vehicle; planning / driving / learning
idm_platoon_simulate	traffic / simulation / vehicle; planning / driving / learning
importance_risk_estimate	sampling / importance / rare; scenario / testing / automated
koschmieder	weather / detection / synthetic
lane_change_permitted	traffic / simulation / vehicle; planning / driving / learning
lateral_wobble	traffic / simulation / vehicle; planning / driving / learning
lidar_scan	traffic / simulation / vehicle; scenario / testing / automated; weather / detection / synthetic; digital / twin / industry
lidar_spec	weather / detection / synthetic
long_simulate	planning / driving / learning; driving / learning / autonomous
ms_ssim	weather / detection / synthetic; driving / learning / autonomous
occlusion_safe_speed	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
osm_road_graph	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
passing_decision	traffic / simulation / vehicle; planning / driving / learning
poisson_events	sampling / importance / rare; scenario / testing / automated
poisson_events_xt	sampling / importance / rare; scenario / testing / automated
project_spherical	weather / detection / synthetic
psnr	weather / detection / synthetic; driving / learning / autonomous
pure_pursuit_curvature	planning / driving / learning; driving / learning / autonomous
range_doppler_map	weather / detection / synthetic
ray_box_ranges	weather / detection / synthetic
ray_plane_range	weather / detection / synthetic
rss_lateral_check	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
rss_longitudinal_check	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
sight_stop_speed	weather / detection / synthetic
skill_test_score	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
social_force_step	traffic / simulation / vehicle; planning / driving / learning
ssim	weather / detection / synthetic; driving / learning / autonomous
ssim_map	weather / detection / synthetic; driving / learning / autonomous
stanley_steer	planning / driving / learning; driving / learning / autonomous
stop_line_plan	planning / driving / learning; driving / learning / autonomous
tdoa_bearing	weather / detection / synthetic
tid2013_evaluate	weather / detection / synthetic; driving / learning / autonomous
tile_stream	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
tof_multipath_phase	weather / detection / synthetic
tof_transport_depth	weather / detection / synthetic
ttc_from_range	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
ttc_truth	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
turn_maneuver_check	scenario / testing / automated; planning / driving / learning; driving / learning / autonomous
veiling_luminance	weather / detection / synthetic
vifp	weather / detection / synthetic; driving / learning / autonomous
world_build	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry
world_camera	traffic / simulation / vehicle; scenario / testing / automated; weather / detection / synthetic; digital / twin / industry; driving / learning / autonomous
world_move	traffic / simulation / vehicle; scenario / testing / automated; digital / twin / industry

This site is open source. Improve this page.