Treffer: Study of a remote object recognition system for an autonomous robot

The goal of the TFG would be to implement an object recognition system for an autonomous robot within the GRASS (Ground Autonomous Student Station) project. The idea is to use two cameras and be able to juxtapose the images and create a 3D vision of what the system is seeing (in our case the rover), and be able to differentiate different predetermined objects by the shape, and if it's possible by the color. The purpose consists in recognizing the position and the orientation of these objects with respect to the rover as the considered reference system. ; The GRound Autonomous Student Station (GRASS) is a student organization with the objective of building an autonomous rover, and one of the capacities needed is the possibility of identifying objects. This project is focused on creating a recognition system capable of being incorporated on the Rover. The developed recognition system is focused on detection by colour and shape, using four colours and one predetermined cylindrical shape. The system has to identify the distance between the object and the camera. The developed system includes a Raspberry Pi system with his camera, using Python and OpenCV libraries as software. The developed system has been tried in a real model, but the final implementation has not been ended due to reasons related with the GRASS project. The structural and logistic parts of the project are developing a new space distribution in order to include eciently the recognition system, although the system has been tested. The developed system is capable of detecting determined objects with a simple shape in a maximum range of 12 meters. At the same time, it works on real environments, counting illumination and environmental visual noise.