Lang, Yilin (Zhejiang University), Li, Zhaoyang (Zhejiang University), Yao, Jinke (Zhejiang University), Shen, Xinwei (Suzhou AiTEN Intelligent Technology Co.,Ltd), Yao, Haijin (Suzhou Aiten Intelligent Technology Co.,Ltd.), Ren, Qinyuan (Zhejiang University)

A Shared Control Approach for Steering Assistance Via Online Learning

Scheduled for presentation during the Regular Session "Driver Assistance Systems I" (FrAT11), Friday, September 27, 2024, 11:10−11:30, Salon 19

2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), September 24- 27, 2024, Edmonton, Canada

This information is tentative and subject to change. Compiled on December 26, 2024

Abstract

Steering assistance systems aim to increase safety and reduce driver workloads. However, inefficient understanding of human driver potential behaviors often leads to increasing human-vehicle conflicts. In this paper, a shared control approach integrating driver behavior online learning is presented for steering assistance. A Gaussian process (GP) is employed for the prediction of the driver future behaviors, which can be continuously updated with real-time driving data. A model predictive controller (MPC) with the GP model is developed to optimize the steering commands with the objective of smooth driving and reducing tracking error. With the trust level of driver behavior prediction, an adaptive authority allocation strategy is adopted to comply with the the change of human-vehicle relationship owing to the online learning. A series of experiments is conducted to verify the proposed method on a lane-keeping task. With the online updating of the driver model and the adaptive authority allocation strategy, the proposed shared control method significantly reduces the tracking errors while enhancing the smoothness of steering simultaneously.