We enhance the navigation capability of a walking robot in unknown terrain by collaboration with a flying robot. In this example the mission is to find and
execute a path from a start area to the goal position. First, a hexacopter with a camera flies over the mission area. During this exploration mission, visual features are tracked for simultaneous localization and mapping. This results in a consistent map of landmarks of the area. At the same time, the camera data is used to create a high resolution elevation map of the environment. In the second part, the walking robot prepares
to navigate with help of the data from the flying robot. The elevation map is analyzed for slope steepness, local roughness, and maximal step height. Based on the combined traversability map our navigation planner finds a path to the goal position. The quadrupedal robot follows the path by using a walking gait while maintaining balance. Accurate state estimation is accomplished by fusing sensor signals from joints and IMU with the localization from image data. The elevation map is continuously updated with the measurements from the laser range sensor. The path is checked for safety and is updated to consider new obstacles, such as this box that we moved in front of the robot. By collaborating with the flying robot, the
walking robot safely navigates through the previously unknown terrain, through narrow passages over the scaffolding, and finally down the ramp to the goal position.