Cooperative Multi-Modal Robots For Autonomous Coverage Path Planning

This paper presents a real-life implementation of a coverage path planning strategy using an Unmanned Ground Vehicle (UGV) and an Unmanned Aerial Vehicle (UAV). The system plans exploration paths for both robots using a wavefront algorithm, accounting for the UGV’s restricted access and the UAV’s limited battery. A hot swapping subroutine allows the UAV to land on the UGV for a manual battery change, enabling continuous exploration. The exploration of a 6.3 × 6.3-meter room took a total of 8.03 minutes, requiring a single battery change after 5.12 minutes of exploration. The UGV took 6.20 minutes, while the UAV needed 1.43 minutes more. Localization is achieved using an indoor localization system with a Kalman filter for the UGV, combining the raw data with local odometry. The unfiltered position of the UAV introduces noise and errors in the estimation that are compensated by the PID controller. Simulations and real-world experiments validate the system, demonstrating the feasibility of heterogeneous robot cooperation.