Hyperspectral Imaging for Vehicle Traction Effort Prediction: ISEAUTO case study

Modern self-driving platforms require new solutions for the more complex scene understanding to develop the driving assistance systems further. This paper presents a new approach for the vehicle traction effort calculation based on the information obtained from a hyperspectral camera. In this case study, the scenarios for different road surfaces are considered to predict the road load affecting the (ISEAUTO) self-driving electric vehicle (EV) designed at Tallinn University of Technology (TalTech). The collected dataset contains the hyperspectral images acquired mainly in the urban environment of the vehicle pathway. Combining the knowledge of the platform parameters and information regarding the road surface material derived from hyperspectral images using a deep learning technique provides a new layer of data that may be used in more accurate and safe vehicle control.