Wang, Weihua (Southeast University), Ding, Haonan (Southeast University), Wang, Ziwei (Southeast University), Yin, Guodong (Southeast University)

Frame-Wise Interactive Strategy on Video Streams for Road Friction Estimation under Kinematic Constraints

Scheduled for presentation during the Invited Session "S21b-Energy-Efficient Connected Mobility" (TH-EA-T21), Thursday, November 20, 2025, 13:50−14:10, Surfers Paradise 3

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

The failure to estimate the road friction coefficient in a timely manner is a major contributing factor to the high incidence of accidents in transportation systems. This paper proposes a reliable road friction estimation method for video streams and designs a frame-wise interactive strategy. The method applies kinematic constraints to image data, optimizing the ROI extraction for real-world driving perspectives, particularly with robustness to trajectory changes during steering. During the estimation phase, we design a multi-patch fusion decision network (MFDNet) for more reliable detection of road surface types. Additionally, considering that real vehicle platforms use video streams rather than 'image-level' data, we design a frame-wise interactive strategy to reduce data redundancy, balance computational resources on the vehicle platform, and mitigate the negative effects of camera defocus and motion blur on road friction estimation. We collected a dataset on closed roads near campus and conducted experiments with the entire algorithm pipeline on a real vehicle. The results show that the proposed method is robust and of practical application value, providing higher safety margins while optimizing the autonomous driving performance of intelligent vehicles.