Vision-based tire deformation and vehicle-bridge contact force measurement

Abstract Accurate measurement of vehicle-bridge contact force is the promise of rapid vehicle-induced bridge impact testing. This paper proposed a cost-effective vertical tire force measurement method based on computer vision and the long and short-term memory (LSTM) neural network. The main contribution of this paper includes two aspects: (1) Circle Hough transform was adopted to monitor tire deformation and vehicle speed from videos of the tire sidewall. Besides, the perspective transformation was employed to eliminate vehicle-induced camera motion; (2) Using tire deformation, vehicle speed, and tire pressure as the input, vertical tire force is estimated through the LSTM neural network. Finally, a detachable vehicle-mounted computer vision system was developed. Experiments, including laboratory and field tests, were conducted. The estimated vertical tire forces show good agreement with the reference values, demonstrating the potential for applying the proposed method to rapid vehicle-induced bridge impact testing.