Estratégias de Navegação para a Inspeção Autônoma de Galerias

Abstract

A critical activity in mining industrial environments is the inspection of regions that present unsanitary working conditions for human professionals, such as caves, pipelines and dam galleries. A possible solution to this problem is the use of a robotic platform for remote inspection. This article focuses on the inspection of galleries with repeated patterns and a limited number of bifurcations. As in this type of task, teleoperation is hampered by several factors such as communication limitations, one solution is autonomous operation. In this context, two autonomous navigation techniques are presented for the exploration of galleries: the first consists in a control by vector fields, while the second is a PID controller implemented with a state machine. Both methods are based on maze solving algorithms, lightweight and easily implemented depending only on a LiDAR sensor. The navigation systems are validated in simulations and with a real robot. In addition, the techniques are compared using two performance metrics, the first one is the total distance traveled and the second one is the absolute distance related to wheel spin. The results obtained hightlight the different characteristics of the navigation strategies according to the inspected environments.