Energy-minimal target retrieval for quadrotor UAVs: trajectory generation and tracking

Abstract

When applying visual servo algorithms for unmanned aerial vehicles (UAVs) using visual registration targets, it is often desirable that the UAV would retrieve targets after the mission. In this study we plan an energy-optimal trajectory for a UAV to retrieve visual registration targets dropped on the initial flight and then construct a controller to track this trajectory in a stable way. The UAV kinematics are described as a Hamiltonian system on the Lie group SE(3); the boundary value problem (BVP) arising from the geometric framing of Pontryagin's Maximum Principle (PMP) is then applied. This BVP is solved using a numerical shooting method with a novel multi-arc interpolation approach determining simple optimal arcs between the targets using a semi-analytical Lie group integrator. A switching controller compensating for the switch between in-ground and out-of-ground dynamics is combined with a disturbance observer to create a controller to track the trajectory in a stable way under the ground effect and bounded disturbances. Simulation results complete the work.