Study, Implementation and Test of a Navigation Architecture for an Autonomous Lunar Nano Drone

The Lunar Nano Drone (LuNaDrone) mission concept was conceived in light of the Artemis Accords signed by Italy and USA in 2020 and is associated with Italy's call to contribute to the development of enabling technologies for lunar surface exploration. The long term vision involves permanent or semi-permanent outposts on the moon, which require new approaches and solutions involving the study of lunar caverns and lava tubes. The goal of this mission is to explore possible entrances to lava tubes via an autonomous spacecraft, for which the Politecnico di Torino has been leading the feasibility and preliminary design studies since early 2020. The focus of this thesis is on the design of the navigation system for LuNaDrone. First, the importance and characteristics of lunar lava tubes, their entrances (skylights), an overview of the discoveries to date and a description of the LuNaDrone mission, spacecraft and flight profile are briefly presented. After studying the requirements for the mission and the state of the art of navigation systems in a similar space application, the object of this thesis becomes the implementation of a visual navigation algorithm in MATLAB in the form of an Extended Kalman Filter. The algorithm makes use of a minimal set of sensors (Camera, IMU and Rangefinder) to operate as a fully self-contained system in difficult environments where no external assistance might be available and within the strict requirements of mass and volume of LuNaDrone. After an introduction to Kalman filters, there follows the description of the mathematical equations necessary for the implementation of the navigation algorithm. These include inertial propagation time update equations and camera and rangefinder measurement update equations involving a hybrid SLAM and MSCKF paradigm, as well as details of state management and vision processing routines. Finally, a prototype of the sensor package along with a mini computer, to record datasets, is developed and built and the data gathered is used for qualitative testing and validation of the algorithm implementation.