Cooperative Control of Swarms of Unmanned Aerial Vehicles

Abstract

Potential function based swarm control is a technique using artificial potential functions to generate steering commands resulting in swarming behavior. This means that the vehicles in the swarm autonomously take care of flying in formation, resulting in steering the swarm, instead of all the individual vehicles being tasked separately. This form of cooperative control is very effective for large groups of unmanned aerial vehicles (UAVs). To test this technique, a simulation tool (in SIMULINK) has been developed using a swarm of quadrotors with nonlinear dynamics. Quadrotors are suitable testbeds due to their mechanical simplicity and can be described by reasonably simple dynamical equations of motion. The inner loop control laws for the quadrotor are based on a backstepping like technique with a nested multi-loop structure. After verification of the individual vehicle model, a multi-UAV controller was designed using potential functions to generate the swarming behavior. The results show that the potential function based method is effective in swarm aggregation. The swarm controller is able to cope with inflight changes of the number of agents if, due to communication limits, one or more agents drift in or out of range. The ability to avoid obstacles by adding obstacle avoidance terms to the steering commands is also shown. These terms are sensitive to the velocities of the swarm members and the gains. However, much better results are expected when a path shaping module is added. This will not only add the ability to fly along a path, but it will also solve the problem of saturated velocity commands due to large differences between the quadrotor position and the goal position.