In this paper a navigation method is presented for space exploration robots using hopping motion in environments with large elevation differences. A monocular camera system is used to reconstruct the flight trajectory and environment around the robot using Structure from Motion while traveling. The created environmental point cloud is projected to 2D to create a variable resolution image and image processing is used to find the most suitable position for the next landing based on normals with the help of gradient maps and error estimation. The method is evaluated in a simulation environment against the previously used protrusion based method to show that the proposed system can extend the operation of the robot to terrains with large elevation differences while still successfully avoid obstacles and dangerous areas.

In this paper a navigation method is presented for space exploration robots using hopping motion in environments with large elevation differences. A monocular camera system is used to reconstruct the flight trajectory and environment around the robot using Structure from Motion while traveling. The created environmental point cloud is projected to 2D to create a variable resolution image and image processing is used to find the most suitable position for the next landing based on normals with the help of gradient maps and error estimation. The method is evaluated in a simulation environment against the previously used protrusion based method to show that the proposed system can extend the operation of the robot to terrains with large elevation differences while still successfully avoid obstacles and dangerous areas.