A Lévy-Inspired Kinodynamic A* Algorithm for Quadrotor Fast Path Searching

Abstract

The advance in research on path searching has enabled quadrotor to navigate autonomously in unknown environments. However, high-speed path searching still remains a significant challenge. Given very limited time, existing path searching methods has no strong guarantee on the feasibility of the solution. This paper proposed a Lévy-inspired kinodynamic A*(LIK-A*) algorithm for quadrotor path searching application. Search strategy with Lévy random walk is introduced in the kinodynamic A* update formula, which helps to increase the perturbation and the diversity of candidate path solution. The search step length is dynamically adjusted during the searching process, making the trajectory to achieve reasonable time allocation and be away from obstacles. To demonstrate the effectiveness of the proposed algorithm, this paper executes simulated flight test. The experiment results show that, comparing to the kinodynamic A* algorithm, the proposed LIK-A* algorithm has better convergence accuracy, stability, and stronger path searching capability, using significantly less memory and time.