Three-dimensional flight experiments using on-line mixed-integer linear programming trajectory optimization

This paper presents an improved path planner formulation using mixed-integer linear programming (MILP) to solve a receding horizon optimization problem in real-time for unmanned aerial vehicles (UAVs). This paper uses variable sized time-steps with linear interpolation points and horizon minimization techniques to enhance the capability of the online path planner. Flight tests on an unique indoor quadrotor testbed demonstrate that MILP can be reliably used in real-time. A multi-vehicle test is also presented to demonstrate non-communicating de-confliction trajectory planning using MILP.