Hierarchical Model Predictive Control for Autonomous Vehicle Area Coverage

Abstract Area coverage using autonomous vehicles receives increasing attention due to a widespread range of possible applications. Examples are surveillance and monitoring tasks or search and rescue missions. Efficient and safe area coverage in dynamic environments, however, is challenging. It requires tight integration of the planning and control task to guarantee collision avoidance and optimal coverage. We propose a combination of two coupled model predictive controllers for optimal area coverage with dynamic obstacle avoidance. The planning is based on a mixed integer programming formulation of the predictive controller. It allows to take dynamic objects, such as other autonomous vehicles into account and considers a simplified dynamic model of the autonomous vehicle. The autonomous vehicle itself is controlled by a continuous time nonlinear model predictive path following controller, which obeys detailed dynamic and kinematic constraints and follows the provided path. The design of the controllers takes the interconnections in terms of dynamic constraints and reference definitions between them into account. Simulation results for a quadcopter illustrate the performance and real-time feasibility of the proposed hierarchical predictive control strategy.