Master of Science

thesisMobile robots are used to venture through types of environments, at low wheel speeds, where wheel slip is a threat. Wheel slip is a hazard to mobile robots in that it introduces error in dead reckoning measurement and in some instances causes the robot to halt its forward progress. To compensate for traction loss several methods are used to determine the terrain characteristics. One of these methods is Pacejka's Tire Model. The slope of Pacejka' s Tire Model can be used to determine when traction loss occurs. One step toward realizing the slope of Pacejka's Tire Model is achieving a good estimate of wheel slip. We present a unique traction estimation algorithm for low speed applications that estimates traction loss by measuring the wheel slip velocity by coupling the dynamics of a wheel with the dynamics of a vehicle. Estimates of the wheel slip velocity are accomplished using onboard sensors. To obtain an accurate estimate of the wheel slip velocity, we propose a modified Kalman Filter that fuses a system model of a DC motor with an estimate of the disturbances acting on the system model. Using the wheel slip velocity a neighborhood can be defined between two instances in time that estimates when traction loss occurs. With means of estimating traction loss, we propose a traction control law for low speed applications that provides the ability of tracking a desired reference while mitigating traction loss. To solve the tracking problem we propose a robust tracking controller that provides the ability of following a defined path and rejecting unmodeled disturbances. To mitigate traction loss we propose a continuous robust traction controller to maximize traction forces by containing wheel slip and its derivative to a neighborhood. The unique aspect of our traction controller is it works jointly with our proposed tracking controller