Safety-Critical Control Allocation for Obstacle Avoidance of Quadrotor Aerial Photography

In quadrotor aerial photography tasks, environmental obstacles may lead to collisions or obstruct the camera's line of sight. Handling the control problems of quadrotor drones under pose constraints has always been a difficult and hot topic. Ensuring compliance with conditions while minimally affecting the system's own control logic and effectiveness is even more challenging. A novel geometry-constructed control barrier function (CBF) is proposed to achieve these challenging goals. Our design leverages geometric features to avoid collisions and line-of-sight obstructions. It overcomes the conservatism of existing CBF, leading to smoother trajectories and smaller corrections to the reference control. Moreover, the quadratic programming (QP) problem of CBF is seamlessly integrated into the control allocation design, which endows the allocator with a safety filter capability when the constraints are activated. This integration does not introduce additional computational complexity and ensures a closer match between actuators' resultant output and the control law's commands compared to separate designs.