Optimization-based Methods for Safe & Robust Aerial Robots
Amr Afifi is a PhD student in the Department of Robotics and Mechatronics. (Co)Promotors are prof.dr.ir. A. Franchi and dr. P. Robuffo Giordano from the Faculty of Electrical Engineering, Mathematics and Computer Science.
This thesis focuses on improving the safety and robustness of optimization-based algorithms for aerial robots. A key challenge lies in the reliance of these algorithms on approximate models of the robot and environment, which can lead to unsafe outcomes when deployed. To address this, the research develops a framework centered on the closed-loop state sensitivity to uncertainty, enhancing the robustness of various optimization-based methods. These include quadratic program whole-body controllers, offline trajectory optimization and model predictive control. By incorporating sensitivity-based robust objectives and constraints into the optimization process, the framework ensures better practical performance and reliability. Extensive testing on aerial robots, both in simulation and real-world experiments, demonstrates the framework’s effectiveness and computational advantages, pushing the boundaries of state-of-the-art safety and robustness in robotic systems.
