---

Under a Phase II SBIR funded jointly by NASA JSC and DARPA, Michigan Aerospace Corporation is currently developing algorithms for processing 3-D range data to extract information about a target spacecraft, such as position, orientation and spin rates. The 3-D sensing and processing concepts are uniquely capable of addressing a wide array of needs in civil space, defense and commercial space communities: from autonomous satellite servicing for extending the life of expensive space assets, docking with assets to transfer orbits or de-orbit, performing space situational awareness for defense purposes, or future Moon and Mars vehicle navigation. Michigan Aerospace Corporation's approach offers the singularly attractive characteristics that it does not require any targeting aids on the asset of interest; it can provide rendezvous at potentially long ranges (kilometers) in addition to close-range docking and inspection; and its solid-state design (no moving parts), minimal size, mass and power, make it practical for micro-satellite platforms. Sensor development and packaging are also being assessed.

Current Software Toolbox and Potential Applications

Software engineers at Michigan Aerospace Corporation are continuously developing and optimizing robust, state-of-the-art machine learning algorithms targeting the unique three-dimensional range data provided by scannerless Laser Detection and Ranging (LADAR) sensors. These 3-D sensor algorithms have been incorporated into a wide variety of cross-platform applications ranging from target satellite pose-determination to instantaneous human gesture recognition. Michigan Aerospace Corporation has an expert team of software engineers dedicated to building an advanced software toolbox for the extraction, analysis and recognition of information and patterns from range images by utilizing the powerful machine learning paradigm provided by ensembles of decision trees. This image-processing toolbox can be applied to many unsolved problems in the scientific, biomedical, military and civilian realms. Some of the potential applications are threat assessment and validation (homeland security/military operations), precise surface damage assessment for air and spacecraft or other objects of interest or structural analysis and fault detection for bridges and similar structures. Michigan Aerospace Corporation is an industry leader in algorithm development and the analysis, and visualization of 3-D sensor data.

Michigan Aerospace Corporation's PointCloudRenderer Software with simulated point cloud
and corresponding range image of the Hubble Space Telescope





PointCloudRenderer displaying data gathered from a LADAR sensor of a in-house 1/24 scale
replica of the Hubble Space Telescope





Michigan Aerospace Corporation's Satellite Pose Determination End-to-End system

This system is designed to predict the pose of the Hubble Space Telescope but can easily be adapted to predict the pose of essentially any well-defined rigid object.