Design of an adaptive neural kinematic controller for a National Instrument mobile robot system

The paper proposes an adaptive neural kinematic controller to guide a National Instrument starter kit robot, which is a nonholonomic differential drive mobile robot, to follow a predefined trajectory. The structure of the controller is based on a neural network topology. The Multi-Layer Perceptron MLP neural network was used to design and implement the controller. The error back propagation algorithm was used to train the neural network to learn the behavior of the mobile robot kinematic model. The neural kinematic controller is trained offline and then the weights of the neural network are adjusted online in order to find the required linear and angular velocities to guide the robot throughout the reference trajectory. The controller is implemented in LabView2011 software and then deployed to the mobile robot platform to allow for autonomous navigation. The simulation results and experimental results show that the proposed controller can successfully navigate the robot along the required path.