Autonomous Quadrotor Trajectory Planning and Control for In-Flight Aerial Vehicle Capture

Quadrotor vehicles provide great flexibility in autonomous flight planning due to their ability for Vertical Take Off and Landing (VTOL) and high maneuverability in the air. This paper leverages these properties to develop a novel path planning and control architecture for an interceptor air vehicle designed to capture a small target drone which is assumed to be at rest or moving slowly. The developed architecture has three main parts. First, a geometric path planner is developed to determine the trajectory for the quadrotor to travel from its initial location to the capture location. The second component of the architecture is a minimal-time thrust profile generation algorithm. The algorithm represents the quadrotor's acceleration as a B-Spline and uses the convex-hull property of the spline to transform the constraints into functions of the control points so an optimization problem can be formulated to minimize time to capture. The third component is a low-level controller that tracks the orientation and thrust commands. In this architecture, the acceleration profile and geometric path are generated independently. This planning and control architecture provides one key advantage over alternative optimal control approaches: by separating the thrust profile generation from the geometric path generation, the minimal time trajectory generation problem has less variables and constraints to calculate, thus allowing on-board calculation of the thrust profile. Results are presented demonstrating aerial capture of targets in simulation. This architecture provides reasonable tracking performance with errors of less than 10 cm when the interceptor air vehicle is traveling at slower speeds and less than 20 cm at higher speeds. It is also shown that the nonlinear optimization problem can be solved in less than 5 ms for on-board calculation.