Quad-Rotor Path Planning for Cluttered and Uncertain Environments

Abstract

This thesis investigates the validity of the path planning algorithm that was developed in a previous study for cluttered and uncertain environments with high-performance and low-performance computing environments.Using Successive Convexification (SCVX) and compound State-Triggered Constraints (STCs), the path planning of a small unmanned aerial system flying through obstacles is assessed. The obstacles are placed with uniform distribution along with the flight course. Various configurations of the obstacles are used in the path-planning computation and the distribution of each computation time and obstacle violations are discussed to assess the path-planning. In evaluating the performance of SCVX, various temporal nodes and maximum number of SCVX iterations are used in the simulation. To assess compound STCs, the dynamics of the vehicle are two quad-rotors connected with a beam-like bar. The simulations are run with both high performance (i.e. normal CPU of a laptop computer) and low performance (i.e. underclocked CPU imitating on-board processor). The results show that the path-planning method is effective and reliable for cluttered and uncertain environments since the computing times converge at certain times that are enough for real-time computation with high-performance computing environments (i.e. normal laptop CPU). In addition, a method that may be able to improve obstacle avoidance and performance with low-performance computing environments is proposed.