Performance comparison of rapidly-exploring random tree algorithms for path planning of autonomous underwater vehicles in complex environments

With the increasing popularity of ocean exploration, research on Autonomous Underwater Vehicles (AUVs) has attracted increased attention. Compared to terrestrial robots, AUVs must endure complex underwater environments and consider various operational factors for path planning missions. In this work, the most used Rapidly-exploring Random Tree (RRT) algorithms, namely basic RRT, RRT*, RRT-connect and Lazy RRT, are implemented and compared for the AUVs’ collision-free path planning in environments with increasing complexity. Path planning performance metrics in terms of Straightness Line Rate (SLR) and Computational Time (CT) are considered under various diving scenarios with increased complexity and dimensionality. Such environments are created based on randomly distributed seamounts and spheres circumscribed obstacles with different dimensions and locations. Demonstrative results are presented, and ANOVA tests are conducted and discussed to compare the path planning performance of the competing RRT algorithms in terms of collision-free navigation capabilities, paths straightness and smoothness, data reproducibility and CPU computation fastness. Through this comparative study, AUVs’ path planning based on RRT* technique operating with a growth process of reselecting parent nodes and reorganizing shows more superiority in terms of itineraries straightness (SLR). The RRT-connect one outperforms all competing RRT algorithms in terms of planning time (CT) and processing fastness.