Simulation of Underwater Environments to Investigate Multi-Robot Systems for Marine Hull Inspection

Abstract

Existing robotics simulations for underwater autonomy systems are limited in many ways, and do not contain enough details to adequately mimic complex inspection operations. In this paper, we work on combining existing tools to create a framework for simulating underwater multi-robot systems for inspection of marine hulls. Using the Robot Operating System (ROS) and Gazebo based platforms, we are able to simulate multiple underwater robots as they complete simultaneous localization and mapping, motion planning, and control. In particular, we demonstrate the capability of our simulation using a new task allocation algorithm based on the consensus-based bundle algorithm, and use this to divide a set of predefined tasks among three robots. Having a simulation with all the tools to run and customize the robots will allow researchers to test and verify new algorithms for underwater inspection without the risks associated with expensive hardware.