Path Planning in a dynamic environment using Spherical Particle Swarm Optimization

TL;DR

This work tackles UAV path planning in dynamic environments with moving threats. It introduces a Dynamic Path Planner based on Spherical Vector-based Particle Swarm Optimization (SPSO-DPP) that re-plans at waypoints using an online SPSO solver. The path is optimized through a multi-objective cost function $F(X_j) = \sum_{k=1}^{4} b_k F_k(X_j)$, incorporating path length, safety, altitude, and smoothness, with the SPSO search conducted in a $3N$-dimensional spherical space via $\Omega_j = (\rho_1, \theta_1, \phi_1, \dots, \rho_N, \theta_N, \phi_N)$. Across four DEM-based scenarios and comparisons against GA and PSO, SPSO-DPP consistently achieves lower costs, with reductions up to $675\%$ relative to GA in the most challenging cases, highlighting its effectiveness for real-time, obstacle-rich UAV navigation.

Abstract

Efficiently planning an Unmanned Aerial Vehicle (UAV) path is crucial, especially in dynamic settings where potential threats are prevalent. A Dynamic Path Planner (DPP) for UAV using the Spherical Vector-based Particle Swarm Optimisation (SPSO) technique is proposed in this study. The UAV is supposed to go from a starting point to an end point through an optimal path according to some flight criteria. Path length, Safety, Attitude and Path Smoothness are all taken into account upon deciding how an optimal path should be. The path is constructed as a set of way-points that stands as re-planning checkpoints. At each path way-point, threats are allowed some constrained random motion, where their exact positions are updated and fed to the SPSO-solver. Four test scenarios are carried out using real digital elevation models. Each test gives different priorities to path length and safety, in order to show how well the SPSO-DPP is capable of generating a safe yet efficient path segments. Finally, a comparison is made to reveal the persistent overall superior performance of the SPSO, in a dynamic environment, over both the Particle Swarm Optimisation (PSO) and the Genetic Algorithm (GA). The methods are compared directly, by averaging costs over multiple runs, and by considering different challenging levels of obstacle motion. SPSO outperformed both PSO and GA, showcasing cost reductions ranging from 330\% to 675\% compared to both algorithms.

Figures (7)

• Figure 1: Environment Isometric & Top Views
• Figure 2: SPSO-DPP Flow Chart
• Figure 3: Base Case: Planned path at each way-point using the same weights as in spso2021
• Figure 4: Safer Path Case: Planned path at each way-point with increasing the value of ${b_2}$ to 100 for ensuring a safer path for the UAV
• Figure 5: Shorter Path Case: Planned path at each way-point with decreasing ${b_2}$ to 1 for prioritizing the distance cost to achieve more dominance in the overall cost function