Xu, Jiahui (Univiersity of Hong Kong), Wang, Yong (Univiersity of Hong Kong), Zhong, Jiaru (Beijing Institute of Technology), Shu, Yiming (The University of Hong Kong), Tang, Jiwei (Univiersity of Hong Kong), Sun, Chen (Univiersity of Hong Kong)

A Third-Party Risk-Aware Decision-Making Method for UAVs in Dynamic Traffic Environments

Scheduled for presentation during the Regular Session "S45a-Decision-Making for Urban Air Mobility and Autonomous Logistics" (FR-LM-T45), Friday, November 21, 2025, 10:30−10:50, Gold Coast

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

Unmanned aerial vehicles (UAVs) are gaining popularity due to their flexibility and agility in modern intelligent transportation system. Ensuring safe decision-making is crucial for UAVs to find suitable paths, especially in dynamic traffic environments. However, most existing research focuses on avoiding collisions with obstacles, with limited attention to risks to third parties, such as crashes and noise pollution. In traffic environments, enabling UAVs to mitigate these risks while making flexible decisions is a key challenge. This paper introduces a risk-aware reinforcement learning method for UAV decision-making in dynamic urban environments with traffic densities and buildings. The method encodes goal and traffic density information, integrating third-party risk factors into the reward function. Static traffic data is processed into dynamic flows for realistic scenario training. Experimental results show that the method effectively manages risks and reduces noise hazards, allowing UAVs to avoid high-density areas while promote safe and environmental friendly operation.