Uncertainty-Aware Motion Planning for Autonomous Driving in Mixed Traffic Environment

2026-06-08

AI Summary

This work addresses the safety risks posed by the high uncertainty of human driving intentions in mixed human–autonomous traffic environments, which often leads deterministic prediction methods to generate unsafe autonomous driving decisions. To tackle this challenge, the paper proposes an Uncertainty-Aware Motion Planning (UAMP) framework that, for the first time, systematically models human intention uncertainty within a reinforcement learning paradigm. Specifically, UAMP employs a proximity-aware uncertainty estimator to construct a joint probabilistic distribution over human intentions under interactive scenarios and introduces an uncertainty-calibrated value learning mechanism. Experimental results demonstrate that UAMP significantly enhances both safety and ride comfort across diverse mixed-traffic scenarios while maintaining traffic efficiency, outperforming existing state-of-the-art methods.

Abstract

In mixed-traffic environments where autonomous and human-driven vehicles may co-exist, motion planning for autonomous vehicles requires anticipating the future behaviors of surrounding human drivers. Existing reinforcement learning-based methods generally directly incorporate the predicted human intents into the observation to enable a proactive planning. However, human intent is inherently uncertain due to the behavioral diversity, perception noise, and partial observability. Treating predicted intends as deterministic states can result in unsafe decisions for autonomous vehicles. To address this problem, we propose Uncertainty-Aware Motion Planning (UAMP), which incorporates uncertainty in human intent prediction for AV decision-making. Specifically, UAMP first introduces a proximity-aware uncertainty estimator to quantify the interaction-conditioned intent uncertainty and constructs an uncertainty-guided joint intent distribution over surrounding human-driven vehicles. Within this uncertainty set, UAMP further introduces Uncertainty-Calibrated Value Learning (UCVL) to correct value function learning biases arising from directly incorporating uncertain human intent predictions into the observation. Extensive experiments in various mixed-traffic scenarios show that UAMP significantly improves safety and driving comfort, while maintaining traffic efficiency compared with existing approaches. The code is released at https://anonymous.4open.science/r/UAMP-5638.

Problem

• uncertainty-aware
• motion planning
• mixed traffic
• human intent prediction
• autonomous driving

Innovation

• Uncertainty-Aware Motion Planning
• Intent Uncertainty Estimation
• Uncertainty-Calibrated Value Learning
• Mixed Traffic
• Autonomous Driving