Uncertainty-Driven Self-Play for Robust Multi-Agent Driving Policies
cs.RO, cs.AI, cs.MA
Submitted: 2026-10-14
Updated: 2026-10-14
Code: https://github.com/Emerge-Lab/PufferDrive
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- nuScenes: A multimodal dataset for autonomous driving
- GraphPilot: Grounded Scene Graph Conditioning for Language-Based Autonomous Driving
- Building reliable sim driving agents by scaling self-play
- CoPark: Learning Reactive Parking via Self-Play
- CounterPlay: Counterfactual Post-Training for Self-Play Driving Policies
- KING: Generating Safety-Critical Driving Scenarios for Robust Imitation via Kinematics Gradients
- Robust Autonomy Emerges from Self-Play
- SaFeR: Safety-Critical Scenario Generation for Autonomous Driving Test via Feasibility-Constrained Token Resampling
- Safety-Critical Traffic Simulation with Adversarial Transfer of Driving Intentions
- Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
- Curiosity-driven Exploration by Self-supervised Prediction
- Large-Scale Study of Curiosity-Driven Learning
- Uncertainty-Aware Reinforcement Learning for Collision Avoidance
- Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles
- Evidential Uncertainty Estimation for Multi-Modal Trajectory Prediction
- FiLM: Visual Reasoning with a General Conditioning Layer
- Large Scale Interactive Motion Forecasting for Autonomous Driving : The Waymo Open Motion Dataset
- NuPlan: A closed-loop ML-based planning benchmark for autonomous vehicles
- PufferLib: Making Reinforcement Learning Libraries and Environments Play Nice
- Proximal Policy Optimization Algorithms
Related papers
- FMT x: An Efficient and Asymptotically Optimal Extension of the Fast Marching Tree for Dynamic Replanning
- MPCFormer: A physics-informed data-driven approach for explainable socially-aware autonomous driving
- RoboLab: A High-Fidelity Simulation Benchmark for Analysis of Task Generalist Policies
- ARC: A Reasoning Recipe for Robot Foundation Models
- HRDexDB: A 4D Dexterous Grasping Dataset Across Human and Multiple Robot Embodiments
- APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies