Uncrewed Vehicle Pathfinding Approach Based on Artificial Bee Colony Method

Abstract

The presented study is dedicated to the dynamic pathfinding problem for UV. Since the automation of UV movement is an important area in many applied domains like robotics, the development of drones, autopilots, and self-learnable platforms, we propose and study a promising approach based on the algorithm of swarm AI. Given the 2D environment with multiple obstacles of rectangular shape, the task is to dynamically calculate a suboptimal path from the starting point to the target. The agent has been represented as UV in 2D space and should find the next optimal movement point from the current position only within a small neighborhood area. This area has been defined as a square region around the current agent’s position. The size of the region has been determined by the attainability of the agent's scanning sensors. If the obstacle is detected by the agent, the latter should be taken into consideration while calculating the next trajectory point. To perform these calculations, the ABC metaheuristic, one of the best representatives of swarm AI, has been used. The validation of the proposed approach has been performed on several 2D maps with different complexity and number of obstacles. Also, to obtain the proper configuration, an inverse problem of identification of guided function weights has been formulated and solved. The outlined results show the perspective of the proposed approach and can complement the existing solutions to the pathfinding problem.