• In this paper, we discuss the design and operation of the prototype of a novel sensor system: HD3D, that fuses high definition LADAR data with synchronized high definition video at frame rate, in near real time. This sensor and the associated algorithms enable proximity sensing and fast estimation of high fidelity geometric models of unknown objects. The range is measured by a MEMs scanning LADAR sensor at a rate of 15 million points/second, within a 1 σ range error of 3 mm, over a 30° field of view. Using an eye-safe realization of the sensor, the present standoff range varies from 1 km down to < 1 m, and a multi-resolution learning algorithm refines the geometry estimates with increasing lateral resolution as the range decreases. Via a priori calibration and synchronization, each impingement point is shown to be mapped into the focal plane pixel address of a high definition video camera. A first generation geometry reconstruction algorithm and its software implementation that enables fusion of overlapping point clouds to establish best estimates of the small body geometry and relative pose of the sensor in near real time is detailed. To this end, a rigorously linear least squares solution is derived for estimation of relative pose parameters to register the point clouds at successive frames. A statistical decision process (using a hypothesis testing procedure from random measurement subsets) is developed to identify consistent measurements while simultaneously obtaining the best motion model. This sensor and algorithm technology is shown to enable highly accurate simultaneous localization and mapping of space objects with high relevance to small body proximity mapping and GN&C. This technology demonstrated using experiments conducted in the Land, Air, and Space Robotics (LASR) laboratory at Texas A&M University.

author list (cited authors)

  • Junkins, J. L., Majji, M., Macomber, B., Davis, J., Doebbler, J., & Nosterk, R.

publication date

  • December 2011