Lunar Autonomy Enabled by Simultaneous Localization and Mapping for Space

NASA seeks Autonomy Enabling Technologies (AETs) in software, mechanical, or combined forms to enhance autonomy, flexibility, and reliability in various space exploration tasks, including vehicle processing, launch operations, and surface maintenance. AETs employ intelligent algorithms and robotic tools crucial for uncrewed systems in extraterrestrial environments. PickNik proposes developing and testing new and existing Visual Simultaneous Localization and Mapping (V-SLAM) algorithms, as well as high-fidelity 3D simulation environments for surface and launch systems operations, to enable the testing and training of robotic surface operations, including excavation and construction tasks. We will upgrade, merge, and link heritage state of the art simulation environments (Gazebo VIPERSim and LunExt) with Space ROS scenarios to allow for interoperable simulation and operation of a variety of different AET tasks, such as automatic traversals from one location to another. This phase 1 focuses on the development of computer vision-based navigation and mapping techniques via simultaneous localization and mapping (SLAM) algorithms. Phase 1 will test and extend ORB-SLAM3, an innovative form of V-SLAM as a key AET for further Lunar development efforts. The rover’s motion model and odometry as well as orientation information from a star tracker system will be added to increase its accuracy and robustness. Additionally, topological features will be labeled via machine learning generated semantic labelings(i.e. rocks, craters, etc) to aid with relocalization and improve map quality. These developments will benefit NASA by improving a variety of autonomous systems including but not limited to mobile based robotic arms, roving systems, and fixed robotic arm systems. Phase 1 will be tested in a software simulation environment, while a future Phase 2 will enhance and mature these capabilities, and include a hardware-based ground demonstration in a mock Lunar environment.