An Autonomous Quadrotor Exploration Combining Frontier and Sampling for Environments with Narrow Entrances
Pudong Liu, Bo Zhang · 2022 41st Chinese Control Conference (CCC) · 2022
An autonomous quadrotor exploration method based on frontier and motion primitives sampling is presented for exploring spatial structural environments with multiple narrow entrances, which is a rigorous challenge for traditional RRT methods. The exploration planning framework includes exploration target selection and autonomous navigation. The local region exploration is implemented by expanding the motion primitives sampling in the exploration target selection, while the global exploration is completed by searching the frontier viewpoints. For local exploration (LE), the motion primitives are expanded according to the quadrotor state, and the best primitive viewpoint is selected by calculating the score of the motion primitives. In global exploration (GE) process, the best global frontier viewpoint is selected, and the key path points are determined in a global undirected graph. After exploring target point is selected, the autonomous navigation module generates a safe, smooth, and dynamically feasible trajectory and enables the quadrotor follow it. The effectiveness and practicality of the exploration framework is validated through experimental simulations, and compared with other existing methods.