Decentralized Control of Unmanned Aerial Vehicle Collaborative Sensing Missions

Cooperative unmanned aerial vehicle (UAV) teams can serve as a mobile sensor networks to autonomously execute sensing tasks in uncertain and dynamic environments. We have implemented a UAV system that performs collaborative sensing missions under the supervision of a single user. Decentralized task allocation and autonomous mission execution are enabled by onboard computing and ad-hoc wireless communication and provide robustness to communication and resource losses in quickly evolving scenarios. The collaboration algorithm combines shared and local information to produce multi-step plans with the goal of minimizing the global cost of the mission. Missions are constructed in real time from tasks such as patrolling an area or searching for an intruder, and may include constraints such as a restricted airspace. Algorithms for decentralized planning, guaranteed search and avoiding a restricted airspace were demonstrated using a team of four UAVs. Experimental data shows that tasks were allocated to the appropriate UAVs and successfully executed.