Some of the most interesting scientific targets for future planetary exploration missions are located in terrain inaccessible to state-of-the-art rover technology, such as exposed impact craters and the Skylight holes on Mars. In order to explore this extreme terrain, the Axel rover has been proposed. It is a two-wheeled tethered robot capable of rappelling steep slopes and traversing rocky terrain. For untethered mobility between scientific targets, two Axel rovers combine and dock with a central module to form the DuAxel rover. Tele-operation of the docking and undocking process of the Axel rovers from the central module has been achieved, but it is a time-consuming process that must be accomplished autonomously for practical space operations. This master’s thesis details the development of a vision-based algorithm to enable autonomous docking and undocking of the Axel rover, to or from the DuAxel rover, following the anchoring of the central module. The algorithm was field tested for performance in the JPL Mars Yard. The proof-of-concept algorithm was able to successfully dock 29 out of 40 tests that were designed to push the algorithm to its limits. Within the limits, the algorithm had an 80% rate of docking successfully. The docking range was found to be 6m and was tested up to 40 degrees, radially centered about the central module, with a relative heading of 20 degrees to the central module. Although many minor improvements can be made, no fundamental challenges preventing autonomous docking were encountered. In an end-to-end demonstration, the algorithm was able to show reliable autonomous docking capabilities for missions in extreme terrain mobility and exploration.