Problem
Research questions and friction points this paper is trying to address.
autonomous driving
multi-agent interaction
risk prioritization
trajectory safety
end-to-end planning
Not All Agents Matter: From Global Attention Dilution to Risk-Prioritized Game Planning
🤖 AI Summary
This work addresses the limitation of existing end-to-end autonomous driving approaches, which treat all traffic participants uniformly and struggle to identify genuine collision risks in complex scenarios. To overcome this, the authors propose GameAD, a risk-prioritized multi-agent game-theoretic framework that formulates driving as a risk-aware interactive decision-making problem. Key innovations include a planning risk exposure metric, risk-aware topological anchoring, a policy payload adapter, a minimax risk-based sparse attention mechanism, and a risk-consistent equilibrium-stabilizing policy. Experiments on the nuScenes and Bench2Drive datasets demonstrate that GameAD significantly outperforms current methods in trajectory safety, validating its robustness and effectiveness in high-risk driving situations.
📝 Abstract
End-to-end autonomous driving resides not in the integration of perception and planning, but rather in the dynamic multi-agent game within a unified representation space. Most existing end-to-end models treat all agents equally, hindering the decoupling of real collision threats from complex backgrounds. To address this issue, We introduce the concept of Risk-Prioritized Game Planning, and propose GameAD, a novel framework that models end-to-end autonomous driving as a risk-aware game problem. The GameAD integrates Risk-Aware Topology Anchoring, Strategic Payload Adapter, Minimax Risk-Aware Sparse Attention, and Risk Consistent Equilibrium Stabilization to enable game theoretic decision making with risk prioritized interactions. We also present the Planning Risk Exposure metric, which quantifies the cumulative risk intensity of planned trajectories over a long horizon for safe autonomous driving. Extensive experiments on the nuScenes and Bench2Drive datasets show that our approach significantly outperforms state-of-the-art methods, especially in terms of trajectory safety.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Risk-Prioritized Game Planning
Minimax Risk-Aware Sparse Attention
Risk-Aware Topology Anchoring
Planning Risk Exposure
End-to-End Autonomous Driving
🔎 Similar Papers
No similar papers found.
