Safe Coverage Control of the Vision-Based UAV System With a Performance-Boosting Strategy

摘要

The visual coverage of a given terrain by a group of unmanned aerial vehicles (UAVs) has found wide applications in various fields, whose focus is on optimizing the coverage quality of multi-UAV system for timely monitoring anomalies in the environment. Nevertheless, the non-convexity of coverage optimization may result in the convergence of multi-UAV system to undesired local optima, and the lack of effective coordination inevitably causes severe safety issues. This technical correspondence aims to address the challenge of elevating the coverage quality of vision-based UAV system with safety constraints. To this end, a metric of coverage cost is designed for region partition by integrating the perception quality with distance information. In addition, a distributed control algorithm with performance boosting strategy is proposed to escape the undesired local optima, thereby improving the coverage performance. Moreover, safety constraints are taken into account in the control design via control barrier functions to secure collision avoidance and flight altitude. Compared to the classic gradient descent method, the coverage performance of the proposed algorithm is improved by more than 40% with a stable coverage quality. With the added benefits of distributed implementation and safety certificates on collision avoidance and flight altitude, it is able to achieve comparable coverage performance to centralized evolutionary algorithms such as particle swarm optimization. Finally, case studies are conducted via numerical simulations and multi-robot experiments to substantiate the efficacy of the safe coverage control approach.