A method for vehicle speed tracking by controlling driving robot

Using a driving robot instead of a human driver to carry out a vehicle emission durability test can effectively improve the accuracy of test. In order to accurately control the speed tracking of driving robot under different working conditions, a speed tracking method of driving robot based on dynamic fuzzy neural network (DFNN) direct inverse control is proposed, which considers the self-learning of vehicle longitudinal performance. Firstly, the kinematics and dynamics models of the driving robot's mechanical legs are established. Moreover, in order to coordinately control the multi-legs of the driving robot to track the vehicle speed, the longitudinal performance model of the controlled vehicle is established by using performance self-learning data and neural network algorithm. Then, to accurately control the movement of the mechanical leg, a direct inverse controller based on DFNN is designed. The output of direct inverse controller is dynamically compensated by closed-loop control of braking force and throttle opening. Finally, the speed tracking experiments and simulations are carried out by human driver, proportional-integral-derivative (PID) control, fuzzy control and the proposed method under different working conditions. The results show that the proposed method does not have the same large prediction error as human driver. At the same time, it can track the speed quickly in different switching conditions. The fluctuation of the tracking speed is small, and the tracking error remains within +/- 1km/h. The proposed method avoids the design of complex control law based on model and can coordinately control the multi-legs to complete the tracking of the target speed.