Implementation of obstacle avoidance algorithms for underwater multi-robot systems.

Abstract

[eng] Navigation of multi-robot systems (MRS) in underwater environments can be challenging due to environmental conditions that can compromise the integrity of autonomous underwater vehicles (AUVs) as a result of possible collisions between them. For this reason, this Master’s Thesis focuses on the design of a Multi-Robot System (MRS) that implements two obstacle avoidance algorithms, with the aim of carrying out numerous 3-D simulations, using the Robot Operating System (ROS) and the Sparus II AUV architecture, in order to obtain results with which to compare the efficiency of both algorithms. Specifically, the efficiency of the classical Artificial Potential Fields (APF) algorithm is compared with the Optimal Reciprocal Collision Avoidance (ORCA) algorithm. Furthermore, the performance of these algorithms when used in combination with a stop-andwait (SAW) strategy is evaluated. The results obtained from the simulations, indicate that, in general, the ORCA algorithm outperforms the APF algorithm in terms of obstacle avoidance efficiency. Furthermore, the work concludes that the SAW strategy is less efficient in terms of navigation time but demonstrates greater robustness in terms of collision avoidance when applied, and tends to yield better results when combined with the APF algorithm. However, conducting numerous simulations is essential to determine the most appropriate navigation strategy based on the environmental characteristics