A Novel MPC Formulation for Dynamic Target Tracking with Increased Area Coverage for Search-and-Rescue Robots

Robots are increasingly deployed for search-and-rescue (SaR), in order to speed up rescuing the victims in the aftermath of disasters. These robots require effective mission planning approaches to determine time and space-efficient trajectories that steer them faster towards (moving) victims, while dealing with uncertainties. Model predictive control (MPC) is an effective optimization-based control approach that has been used to steer robots along reference trajectories determined by higher level controllers. Determining the trajectory of the robots directly via MPC has the advantage of optimizing multiple SaR criteria while handling the constraints. We, thus, introduce a path planning approach based on MPC for indoor SaR robots that allows the robot to systematically chase the moving victims, when no reference trajectory is provided. The proposed approach combines target-oriented and coverage-oriented search, and allows for systematic handling of environmental uncertainties, by deploying a robust tube-based version of the introduced MPC formulation. In addition, we model the movements of the victims for MPC, by adopting an existing evacuation model. We present a case study, using Gazebo, MATLAB, and ROS, where the performance of the proposed MPC controller is evaluated compared to four state-of-the-art methods (two target-oriented methods based on MPC and A* and two heuristic algorithms for area coverage). The results show that, while robust to uncertainties, our approach overall outperforms the other methods, with regards to victim detection, area coverage, and mission time.