Artificial Neural Network (ANN)-Based Supervised Control of Single Wheel Robotic System (SWRS)

This study considers a supervised learning technique based on feedforward artificial neural network (ANN) for control of highly nonlinear and multivariable single-wheel robotic system (SWRS). The nonlinear governing equations of motion for the system were derived using Newton’s second law and simulated in MATLAB/Simulink platform. The outcomes of PID control were considered for training of ANN controller. Excellent regression results of almost 0.99 were obtained after training. Finally, a comparison analysis has been done between PID and ANN controllers in terms of settling time, overshoot ranges and steady state error response.