Joint Level Adaptive Control for Mobile Robots in ROS 2

Abstract

This paper proposes an adaptive controller for mobile robots using the navigation framework of the Robot Operating System. The navigation framework implements a kinematic controller which considers kinematic constraints such as robot nonholonomic constraints and obstacles and outputs linear and angular velocity commands to be implemented by the joint level controller. Usually the joint level relies on velocity servos, which neglect the dynamics of the robot or PID controllers, which are difficult to tune, due to the nonlinear model and unknown parameters of the model. The proposed model reference adaptive controller accepts velocities references, as required by the navigation framework and outputs the wheel torques to impose the commanded velocities. A formal proof of stability including the robot dynamics, controller and adaptation law is provided. Although it is not proved, as it depends on persistence of excitation on the reference signal, experimental results shows that the estimated parameters converge to values close to the actual ones.