A Simulated Environment for the Development and Validation of an Inspection Robot for Confined Spaces

Abstract

Some of the most demanding tasks performed in industrial environments are inspections and mapping, particularly in pipes, caves, and dam galleries. Given the nature of the manual inspection process, these confined spaces environments present hazardous conditions for human inspectors. In this sense, a mining company created an inspection robot for confined environments to reduce the operator's risk. With a focus on improving the validation of new navigation and control algorithms, different mechanical proposals, and efficient remote operator training, this work presents a methodology for creating a simulated environment for inspection robots. We present the pipeline for realistic virtual environment generation using an adapted robot model, including lifelike map generation methods such as photogrammetry. Performance evaluations show the efficiency of the parameters and models used in the simulation. Finally, we present mapping, path planning, and control results using the simulated robot, which can also be executed directly on the real robot.