Sistema de estimación de transitabilidad para robots móviles terrestres

Abstract

This project presents the development of a robotic perception system for estimating transitability in external and unstructured environments, as well as its subsequent implementation in the land mobile robot Jackal. Similarly, this project is framed in one of the lines drawn by the Semillero de Investigación en Robótica de la Universidad Autónoma de Occidente (UAO), in order to support forest characterization tasks. The system presented here allows to estimate transitability, through different techniques, based on the detection of obstacles in an explicit way, to later carry out the rethinking of the route defined at the beginning and, finally, its execution by Jackal. It is important to mention that, for the development of the same one, an intense search of the state of the art was made around the systems of estimation of transitability, then a sifting of the alternatives, combination (or addition of libraries like PCL or Costmap2d), the re-parametrization of the same ones and the use of sensors LIDAR, stereo camera, GPS and IMU, arriving at more robust solutions. Finally, physical tests were carried out to determine the correct functioning of the same; it should be clarified that ROS was used as middleware with the support of the C++ programming language. In addition, these tests were performed in two stages: inside the robotics laboratory of the UAO and outside it (external and unstructured environments), the latter being the most important for the present project, managing to determine the proper functioning of the algorithms, as long as the parameters are established correctly.