Attention and Risk-Aware Decision Framework for Safe Autonomous Driving

2025-09-09

AI Summary

To address the low training efficiency and high collision risk of Proximal Policy Optimization (PPO) in long-horizon autonomous driving tasks, this paper proposes a risk-aware enhanced attentional decision-making framework. Methodologically, it integrates a dual-path channel-spatial attention mechanism to selectively focus on high-risk regions in visual inputs, designs a risk-attention network coupled with a dynamically balanced reward function, and incorporates a real-time safety-assistive supervision module. The key contribution lies in explicitly embedding risk awareness into the policy optimization process, enabling precise hazard identification and proactive intervention. Extensive experiments across diverse traffic-flow scenarios demonstrate that, compared to baseline PPO, the proposed method reduces peak reward acquisition time by 32%, accelerates training convergence by 27%, improves obstacle avoidance success rate by 19.5%, and decreases traversal time in high-risk zones by 41%, thereby significantly enhancing both safety and training stability.

Technology Category

• Search and Optimization: Learning to Search
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Autonomous driving has attracted great interest due to its potential capability in full-unsupervised driving. Model-based and learning-based methods are widely used in autonomous driving. Model-based methods rely on pre-defined models of the environment and may struggle with unforeseen events. Proximal policy optimization (PPO), an advanced learning-based method, can adapt to the above limits by learning from interactions with the environment. However, existing PPO faces challenges with poor training results, and low training efficiency in long sequences. Moreover, the poor training results are equivalent to collisions in driving tasks. To solve these issues, this paper develops an improved PPO by introducing the risk-aware mechanism, a risk-attention decision network, a balanced reward function, and a safety-assisted mechanism. The risk-aware mechanism focuses on highlighting areas with potential collisions, facilitating safe-driving learning of the PPO. The balanced reward function adjusts rewards based on the number of surrounding vehicles, promoting efficient exploration of the control strategy during training. Additionally, the risk-attention network enhances the PPO to hold channel and spatial attention for the high-risk areas of input images. Moreover, the safety-assisted mechanism supervises and prevents the actions with risks of collisions during the lane keeping and lane changing. Simulation results on a physical engine demonstrate that the proposed algorithm outperforms benchmark algorithms in collision avoidance, achieving higher peak reward with less training time, and shorter driving time remaining on the risky areas among multiple testing traffic flow scenarios.

Problem

• Improving PPO algorithm training efficiency and safety in autonomous driving
• Addressing collision risks during lane keeping and changing maneuvers
• Enhancing decision-making for unforeseen events in complex traffic scenarios

Innovation

• Risk-aware mechanism highlights potential collision areas
• Balanced reward function adjusts for vehicle density
• Safety-assisted mechanism prevents risky lane maneuvers