Jump-and-flight locomotion for a hybrid aerial loco-manipulator using MPC

Abstract

This thesis investigates the trajectory generation and control of agile locomotion for aerial manipulators by using optimal control for trajectory optimization and model predictive control (MPC) for real-time execution. The study formulates robot tasks as optimal control problems constrained by full-body dynamics that explore agile and dynamically feasible trajectories. These trajectories are performed through an MPC controller that solves the problem iteratively and applies the resulting commands in a closed-loop manner. To ensure real-time feasibility, the work employs a solver based on differential dynamic programming, with specific techniques to handle actuator bounds. The proposed approach is validated through simulations and experiments on an aerial manipulator with a hexarotor platform. The results demonstrate that combining trajectory optimization with MPC enables agile locomotion that would be difficult to achieve using traditional methods.