UAVs Path Planning by Particle Swarm Optimization Based on Visual-SLAM Algorithm

Abstract

AbstractIntelligent 3-D path planning is a crucial aspect of an unmanned aerial vehicle's (UAVs) autonomous flight system. In this chapter, we propose a two-step centralized system for developing a 3-D path-planning for a swarm of UAVs. We trace the UAV position while simultaneously constructing an incremental and progressive map of the environment using visual simultaneous localization and mapping (V-SLAM) method. We introduce a corner-edge points matching mechanism for stabilizing the V-SLAM system in the least extracted map points. In this instance, a single UAV performs the function using monocular vision for mapping an area of interest. We use the particle swarm optimization (PSO) algorithm to optimize paths for multi-UAVs. We also propose a path updating mechanism based on region sensitivity (RS) to avoid sensitive areas if any hazardous events are detected during the execution of the final path. Moreover, the dynamic fitness function (DFF) is developed to evaluate path planning performance while considering various optimization parameters such as flight risk estimation, energy consumption, and operation completion time. This system achieves high fitness value and safely arrives at the destination while avoiding collisions and restricted areas, which validates the efficiency of proposed PSO-VSLAM system as demonstrated by simulation results.KeywordsVisual-SLAMPSOPath planningAutonomous aerial vehiclesUAV