Basic Study on Driving Force Control for EVs Using Road Condition Estimation with Machine Learning Image Recognition and Vehicle Dynamics

Driving Force Control (DFC) for in-wheel motor electric vehicles has been studied for years. To realize DFC, setting the slip ratio limiter (SRL) through appropriate estimation of the road-tire model is essential on low friction surface. However setting the slip ratio limiter based on vehicle dynamics has faced the issue of tire model estimation delay, a fixed SRL has been commonly used in the traditional DFC. This paper proposes a method to quickly adapt to changes in road conditions by the fusion of vision based SRL estimation with vehicle dynamics based SRL estimation. At the moment of road surface condition transition, the vision-based estimation is referenced to compensate for the delay in the dynamics-based approach using fusion algorithm. Experiments and simulations confirmed an increase in traction force on low friction roads through appropriate setting of slip ratio limiter. Experimental results demonstrated that immediately after a sudden change of road surface condition, the proposed method increased driving force by 15 % compared to conventional DFC systems with a variable SRL based solely on dynamics-based estimation.