Human Rescue Based On Autonomous Robot Kuka Youbot With Ros And Object Detection

The rescue processes that must be carried out in environments that present contamination levels are limited by the exposure time to which rescuers can be submitted, due to this the use of robotic platforms has been taken as a feasible option since it allows to increase the precision when carrying out the operations and to reduce the time in the decision making. In this paper is presented a system which uses the Kuka Youbot mobile platform to perform navigation processes, exploration and detection of people, in order to generate a map which contains the locations of individuals detected. The processes are executed through the combination of Robot Operating System and Open Source Computer Vision Library, in addition the complete system is simulated within the environment of Gazebo. The detection of individuals is done by implementing the Single Shot Detector algorithm and using a trained neural network. The simulation of the system made it possible to detect people inside a building in spite of occlusions in the captured images and it was possible to save the coordinates of their locations to later generate marks inside a 2D generated map.