Ph.D. Defense

Mehregan Dor 

(Advisor: Prof. Panagiotis Tsiotras)

"Autonomous and Robust Monocular Simultaneous Localization and Mapping-Based Navigation for Robotic Operations in Space"

Monday, April 22 

1:00 p.m. 

Montgomery Knight Building 325

Abstract 

Precise relative navigation techniques, incorporating increased levels of autonomy, will be a key enabling element of future missions involving spacecraft Rendez-vous and Proximity Operations (RPO) or Small-Body Probing and Surveying (SBPS). It is typically recognized that the high-risk nature of missions with proximity operations, along with a lack of autonomy in current mission procedures, severely limits the possibilities in mission design. Indeed, ground-segment operators are intimately involved in all in-situ tasks, which ultimately rely on extensive human-in-the-loop verification, as well as ground-based computations for estimation, guidance, and control. In addition, long round-trip light times and severely limited bitrate in communications render ground-in-the-loop processes extremely tedious. Nevertheless, safety critical maneuvers in both RPO and SBPS require on-board closed loop control predicated on precise relative navigation. In tandem, we expect that the incorporation of autonomous capabilities has the potential to improve navigation performance and reduce operational complexity for future missions. In line with the drive to improve the autonomy of such processes, we explore and apply methods typically used in ground robotics to real-world spacecraft relative navigation problems. In past and current space missions of this nature, the severe lack of computational resources has led some authors to consider filtering as the only adapted space-bound navigation solution, generally exploiting a combination of bearing and range measurements in some way to produce navigation solution. Given the ever growing on-board computer resources available in space-bound missions however, large scale graph-based Simultaneous Localization and Mapping (SLAM) approaches, combining monocular camera measurements with other sensor modalities, may soon enough be flown as viable solutions for on-the-fly autonomous relative navigation in unmanned space missions, such as on-orbit satellite servicing and large structure assembly, or deep-space operations in the vicinity of small bodies. Such systems enable navigation and closed-loop control of the chaser's motion around the target object, intended for specific safety critical maneuvers. In this thesis, the author first identifies real-world challenges related to the application of SLAM to the spacecraft relative navigation problem --- by leveraging real RPO mission image sequences. Then, orbital motion constraints specific to the small-body circumnavigation problem are modelled and incorporated in to make the SLAM solution more robust to outliers and drift in the form of the new factor, predicated on the relative motion constraints, as opposed to the typical inertial sensor approach. Further details on the implementation of AstroSLAM, an algorithm permitting navigation around small-celestial bodies, fusing monocular camera and other sensor measurements to perform spacecraft-target relative state estimation, are provided. Validation of

AstroSLAM is carried out using data from previously flown missions and data generated in-lab. The work lays out and demonstrates with results how to perform on-the-fly parameter estimation (such as spin state and gravity) for the SBPS problem. The challenges related to SLAM initialization arising in the spacecraft navigation application specifically, suffering from weak-perspective projection and small baselines, are also tackled by means of incorporating recent a methodology in ground-based robotics.

Committee

· Prof. Panagiotis Tsiotras – School of Aerospace Engineering (advisor)

· Prof. Glen Lightsey– School of Aerospace Engineering

· Prof. John Christian – School of Aerospace Engineering

· Prof. Frank Dellaert – School of Interactive Computing, CTO Verdant Robotics

· Prof. Luca Carlone – Department of Aeronautics and Astronautics, MIT