Point Cloud Based Relative Navigation for Small Satellite Rendezvous

Abstract

The thesis describes the development and testing of a point cloud based relative navigation sensor. The objective of this research is to provide a low cost, lightweight, low power, and low volume alternative to established rendezvous sensors , in order to enable small satellites to conduct missions such as in-space inspection, on-orbit servicing, and space debris removal. The sensor system must be able to provide relative position and orientation data to a control system independent of the cooperation of the target spacecraft or object. This is done using commercially available hardware (Microsoft Kinect v2) and an open source C++ library (Open3D). The thesis details the logic of the software implementation and provides data to validate the accuracy and precision of the system. Based on lab tests in a controlled environment, the system is able to measure relative position to within 0.005 m along its boresight axis at a rate on the order of 1 Hz. It does so without any knowledge of the target geometry and without specially designed navigation aids placed on the target. The error is based on a “ground truth” position as measured by a 12 camera OptiTrack system which has sub millimeter accuracy and precision.