This master thesis presents a method for 3D navigation of a robotic platform in urban environments. In autonomous navigation, the robot must know the localization of all the environment obstacles, so an algorithm for obstacle detection is developed and tested using a LiDAR and a camera as sensors, comparing the data points’ height. This detection focus on objects the robot could collide with in urban environments, including negative obstacles such as holes or stairs. The navigation and detection algorithms are all integrated in ROS (Robot Operating System). The simulation and experimental results show the effectiveness of the algorithm to detect those obstacles, being successful with the LiDAR as a sensor in urban environments, but not sufficient robust enough for the camera when the navigation is done outdoors with high sunlight.