Final Report - 484 (Final Report)
- Download File
- Final_Report_-_484.pdf
- Description
- This research aims to contribute to the state-of-the-art of current research on VRU safety. Following are the three main contributions of this research: (1) Developing a disturbance input type communication-disturbance-observer (CDOB) based delay-tolerant control strategy to enhance robustness of the control under computation latency, network latency and packet loss. The aim here is to improve the performance of collision avoidance path maneuvering in real life situations with inherent delays in the loop. (2) Developing a hierarchical control framework that integrates deep reinforcement learning (DRL) for high-level decision-making with a CLF-CBF-QP-based controller for safe and smooth low-level execution, with application to VRU safety. (3) Introducing a novel VVE-based testing pipeline that enables efficient and rigorous evaluation of autonomous driving functions under various complex and high-risk traffic scenarios with specific application to VRU safety.
- Citation
- Publication Date
- July 31, 2025