Perception and navigation in unknown environments: The DARPA Robotics challenge

Abstract

This paper presents different techniques to achieve the tasks proposed in the DARPA (Defense Advanced Research Projects Agency) VRC (Virtual Robotics Challenge), which entails the recognition of objects, the robot localization and the mapping of the simulated environments in the Challenge. Data acquisition relies on several sensors such as a stereo camera, a 2D laser, an IMU (Inertial Motion Unit) and stress sensors. Using the map and the position of the robot inside it, we propose a safe path planning to navigate through the environment using an Atlas humanoid robot.