Relative Vehicle Displacement Approach for Path Tracking Adaptive Controller With Multisampling Data Transmission

This paper proposes a cyber-physical framework for vision-based automated vehicle path tracking. The framework has been developed in two parts: first, an intelligent relative vehicle displacement approach with an adaptive neural network controller; and second, multisampling data transmission architecture for any sensor-controller-actuator network. Uncertainties due to illumination effects, occlusion, and obscure images affect system performance drastically. The proposed relative vehicle displacement approach takes care of these uncertainties. The adaptive neural network controller generates precise control actions for stabilizing the system in minimum time. A reliable and robust data transmission architecture is of utmost importance for any Internet of Things application. Successful data transmission depends on several parameters, such as delay, communication channel behavior, and packet loss. A novel multisampling data transmission architecture addressing these issues has been proposed in this paper. Various cases of data transmission have been demonstrated on the time critical application of vision-based tracking by an automated guided vehicle. The results of real-time path tracking operation have been duly compared with other control techniques and data transmission architecture. The experimental results show the efficiency of the proposed system.