Autonomous Proximity Operations for Noncooperative Space Targets

摘要

Future space debris removal and On-Orbit Servicing missions to refuel, repair or deorbit end-of-life and defunct satellites require proximity opera-tions and, eventually, docking to non-cooperative space vehicles. The main challenge in these operations is due to the non-cooperative nature of the target, since it is not equipped with visual markers for attitude and dis-tance (i.e. pose) estimation, it does not communicate with the chaser and it may rotate or tumble about the principal axes. Existing state-of-the-art robotic systems performed proximity operations and docking only to either cooperative or partially cooperative attitude-stabilized targets in test mis-sions after overcoming contingencies through ground support. The capabil-ity to autonomously rendezvous and dock to noncooperative space targets was considered by NASA a high-priority cross-cutting technology required for the decade 2011-2021 and beyond. Autonomous systems would be able to service the majority of space debris and end-of-life satellites using the same algorithms and methods. Therefore, most servicing missions can be performed with the same kind of technology. This would ensure frequent On-Orbit Servicing missions. This research is motivated by the hypothesis that it is possible to develop a complete Guidance, Navigation and Control system capable to perform autonomous proximity operations for non-cooperative space targets. To investigate methods for safe and reliable proximity operations to noncoop-erative targets, an innovative Guidance, Navigation and Control system is developed, implemented and verified step-by-step in this thesis. The first part of the research develops the guidance algorithm for onboard generation of the approach trajectory. The dynamics model for proximity operations between a controlled chaser spacecraft and an uncontrolled tar-get is first defined. The dynamics model includes the path constraints re-quired for safe maneuvering. As a particular test case, the docking maneu-ver is considered. Based on this model and the requirements for the Guid-ance, Navigation and Control system, a methodology for rapid trajectory generation is chosen among state-of-the-art methods. The key feature of the selected algorithm is that the trajectory of the chaser spacecraft is pa-rameterized by a set of polynomials. Using the dynamics model of the ma-neuver and a nonlinear programming solver, the coefficients of polynomials are optimized while respecting the path constraints. Due to the limited number of variables, the optimization of the 6 degrees-of-freedom trajecto-ry can be executed in between 0.5 s and 9.0 s, allowing onboard execution. Optimality, maneuvering precision and robustness to orbital perturbations and sensor noise are evaluated through extensive numerical experiments. The second part of the research focuses on the navigation system. The al-gorithms, including the navigation filter, for model-based, noncooperative pose estimation using data from Lidar, stereo camera and mono camera are developed separately. These sensors are considered since they are state-of-the-art systems in terrestrial robotics. The performance of the proposed methods are evaluated through an experimental campaign using the RA-COON laboratory, which provides real-time full degrees-of-freedom hard-ware simulation of proximity operations under realistic lighting conditions. Test results lead to the choice of the most suitable rendezvous sensor and related attitude estimation technique that complete the innovative Guid-ance, Navigation and Control system.