Fast Planner for MAV Navigation in Unknown Environments Based on Adaptive Search of Safe Look-Ahead Poses

Abstract

Autonomous navigation capability is a crucial part for deploying robots in an unknown environment. In this article a reactive local planner for autonomous and safe navigation in subterranean environment is presented. The proposed planning framework navigates the MAV forward in a tunnel such that the MAV gains more information about the environment while avoiding obstacles. The proposed planning architecture works solely based on the information of local surrounding of the MAV thus, making navigation simple yet fast. One of the biggest novelties of the article comes from solving the combined problem of autonomous navigation and obstacle avoidance. The proposed algorithm for selecting the next way point of interest also accounts in the safety margin for traversing to such way point. The approach presented in this article is also different from classical map based global planning algorithms because it favours the next way point away from obstacles in way point selection process and thus providing a safe path for incremental forward navigation. The approach is validated by simulating a MAV equipped with the proposed reactive local planner in order for the MAV to navigate in a subterranean cave environment.