Fail2Drive is the first paired-route benchmark for closed-loop generalization in CARLA, showing an average 22.8% success-rate drop on shifted scenarios and revealing failure modes such as ignoring visible LiDAR objects.
hub
Robust autonomy emerges from self-play.arXiv preprint arXiv:2502.03349
14 Pith papers cite this work. Polarity classification is still indexing.
14
Pith papers citing it
hub tools
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
Reward design in Perceiver-based RL driving agents determines attention allocation to elements like GPS tokens and risks, inducing a learned vigilance prior and sometimes reversing attention-risk correlations, with a statistical correction for measuring these links.
Self-play in a vector simulator trains a driving teacher; a vision student aligned by action and low-rank structural losses drives end-to-end without expert trajectory supervision.
A hierarchical Stackelberg MARL plus continuous-motion architecture with hybrid co-training produces smoother and safer closed-loop traffic behavior than standard self-play methods.
Closed-loop on-policy training with a reactive goal-oriented scene decoder cuts collision rates by up to 79.5% in dense traffic compared to standard open-loop baselines.
Pseudo-expert regularized offline RL reduces collisions and improves route completion for camera-based driving models trained on fixed simulator datasets from nuScenes.
SimScale synthesizes unseen driving states from real logs via neural rendering and reactive environments, generates pseudo-expert trajectories, and shows that co-training on real plus simulated data improves planning robustness and generalization on real benchmarks, with gains scaling by simulation
ReSim is a controllable video world model trained on heterogeneous real and simulated driving data that achieves higher fidelity and controllability for both expert and non-expert actions, plus a Video2Reward module for estimating action quality from simulated futures.
Genetic algorithm optimizes parameters of multi-agent flocking models to match user-defined objectives, with alignment emerging from spacing maintenance.
Self-play RL regularized with 30 minutes of human data produces driving policies that coordinate with humans, training in 15 hours on one GPU with 2500x less data than imitation learning.
A C++ Dec-POMDP simulator using data-oriented design and zero-copy PyTorch integration achieves up to 33 million steps per second on a 16-core CPU, enabling multi-agent policy training in minutes with PPO, DQN, and SAC.
This survey synthesizes AI techniques for mixed autonomy traffic simulation and introduces a taxonomy spanning agent-level behavior models, environment-level methods, and cognitive/physics-informed approaches.
An instance-centric representation with local frames, relative positional encodings, and adaptive reward transformation in adversarial IRL yields scalable, accurate, and robust behavior models for multi-agent driving simulation.
PODS applies max-variance down-sampling to GRPO rollouts in LLM RLVR, delivering at least 1.7x faster training to peak test accuracy on reasoning benchmarks.
citing papers explorer
-
Fail2Drive: Benchmarking Closed-Loop Driving Generalization
Fail2Drive is the first paired-route benchmark for closed-loop generalization in CARLA, showing an average 22.8% success-rate drop on shifted scenarios and revealing failure modes such as ignoring visible LiDAR objects.
-
Reward-Conditioned Attention: How Reward Design Shapes What Autonomous Driving Agents See
Reward design in Perceiver-based RL driving agents determines attention allocation to elements like GPS tokens and risks, inducing a learned vigilance prior and sometimes reversing attention-risk correlations, with a statistical correction for measuring these links.
-
TerraTransfer: Learning End-to-End Driving Policies Without Expert Demonstrations
Self-play in a vector simulator trains a driving teacher; a vision student aligned by action and low-rank structural losses drives end-to-end without expert trajectory supervision.
-
Beyond Self-Play: Hierarchical Reasoning for Continuous Motion in Closed-Loop Traffic Simulation
A hierarchical Stackelberg MARL plus continuous-motion architecture with hybrid co-training produces smoother and safer closed-loop traffic behavior than standard self-play methods.
-
Goal-Oriented Reactive Simulation for Closed-Loop Trajectory Prediction
Closed-loop on-policy training with a reactive goal-oriented scene decoder cuts collision rates by up to 79.5% in dense traffic compared to standard open-loop baselines.
-
Pseudo-Expert Regularized Offline RL for End-to-End Autonomous Driving in Photorealistic Closed-Loop Environments
Pseudo-expert regularized offline RL reduces collisions and improves route completion for camera-based driving models trained on fixed simulator datasets from nuScenes.
-
SimScale: Learning to Drive via Real-World Simulation at Scale
SimScale synthesizes unseen driving states from real logs via neural rendering and reactive environments, generates pseudo-expert trajectories, and shows that co-training on real plus simulated data improves planning robustness and generalization on real benchmarks, with gains scaling by simulation
-
ReSim: Reliable World Simulation for Autonomous Driving
ReSim is a controllable video world model trained on heterogeneous real and simulated driving data that achieves higher fidelity and controllability for both expert and non-expert actions, plus a Video2Reward module for estimating action quality from simulated futures.
-
EvoFlock: evolved inverse design of multi-agent motion
Genetic algorithm optimizes parameters of multi-agent flocking models to match user-defined objectives, with alignment emerging from spacing maintenance.
-
Human-like autonomy emerges from self-play and a pinch of human data
Self-play RL regularized with 30 minutes of human data produces driving policies that coordinate with humans, training in 15 hours on one GPU with 2500x less data than imitation learning.
-
A High-Throughput Compute-Efficient POMDP Hide-And-Seek-Engine (HASE) for Multi-Agent Operations
A C++ Dec-POMDP simulator using data-oriented design and zero-copy PyTorch integration achieves up to 33 million steps per second on a 16-core CPU, enabling multi-agent policy training in minutes with PPO, DQN, and SAC.
-
Artificial Intelligence for Modeling and Simulation of Mixed Automated and Human Traffic
This survey synthesizes AI techniques for mixed autonomy traffic simulation and introduces a taxonomy spanning agent-level behavior models, environment-level methods, and cognitive/physics-informed approaches.
-
Toward Efficient and Robust Behavior Models for Multi-Agent Driving Simulation
An instance-centric representation with local frames, relative positional encodings, and adaptive reward transformation in adversarial IRL yields scalable, accurate, and robust behavior models for multi-agent driving simulation.
-
Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning
PODS applies max-variance down-sampling to GRPO rollouts in LLM RLVR, delivering at least 1.7x faster training to peak test accuracy on reasoning benchmarks.