Markov Decision Process-Based Collision Avoidance Method for Multirotor Small Unmanned Aircraft System

The utilization of small unmanned aircraft systems (sUAS), also known as drones, is growing as their performance improves. In particular, multirotor sUAS are expected to have wide-ranging applications, such as delivering goods and aerial surveillance, but the rising number of sUAS flights is also exacerbating the risk of collision between sUAS and manned aircraft. This study develops a collision avoidance method for multirotor sUAS using the Markov decision process (MDP). Recently, the MDP-based collision avoidance method has been developed for various kinds of aircraft including sUAS. However, the collision avoidance method for sUAS is developed without distinguishing multirotor and fixed-wing sUASs from a compatibility perspective. We propose a collision avoidance method that optimally exploits the advantageous flight characteristics of multirotor sUAS, such as their ability to change acceleration dynamically and hover. Numerical simulations are conducted to demonstrate the performance of the proposed method. The results show that the proposed collision avoidance method can avoid collisions while suppressing excess deviation from the original path.