PROJECTS

In the last decade, disaster-response robotics has gained more and more attention in the robotic community for its potential benefit in supporting rescuers on disaster sites.  A crucial problem for such robots, which is barely addressed in the literature, is the impact of thermal fatigue, which may deteriorate the task performance and eventually damage the robot. This thesis proposes to face the problem from a Nonlinear Model Predictive Control (NMPC) perspective, such that the motor thermal model is directly exploited to predict and constrain the motor temperature.

 

Based on optimal control problem, this work shows how the nonlinear model predictive control algorithm is implemented in the Robotic Operating System (ROS) environment. The control algorithm is finally tested on the CENTAURO platform by performing two heavy manipulation tasks, ensuring the prevention of thermal burnout. The proposed NMPC effectively prevented the thermal burnout by adapting the robot configuration at run-time, finding a trade-off between task completion and fulfillment of the thermal fatigue constraint.