Edward Gregson, Dalhousie University; Mae Seto, Dalhousie University; Bumsoo Kim, Defence R&D Canada
Keywords: Space Debris, collision avoidance, autonomy
Abstract:
This paper reports on a study to assess the value of on-board-the-satellite robotic autonomy towards satellite collision avoidance against space debris or other satellites. While currently ground control station operators are heavily involved in collision avoidance maneuver planning, the work reported proposes that autonomy techniques can be applied to lessen their required role, and make the satellite more robust to unforeseen situations. The advantages include reduced workload for ground control staff and more flexibility in the responses to a wider variety of collision avoidance situations. The autonomy includes optimization, decision-making, mission re-planning, learning, etc. It will be designed to work with the lists of two-line elements (TLEs) released by US authorities. The collision avoidance autonomy can be implemented in one of 3 ways: 1. the autonomy resides solely on-board the satellite; 2. the autonomy is in the form of tools that ground control engineers apply to automate or assist with the decision-making, and 3. The autonomy is distributed in a hybrid manner between the satellite and ground control engineers. The paper provides a brief literature review on the state-of-the-art of autonomy on satellite and spacecraft control in general. Then, the paper examines what could reasonably be achieved with autonomous control for satellite collision avoidance. To assess the value of autonomy for spacecraft collision avoidance, a research testbed was developed in the Intelligent Systems Laboratory, Dalhousie University to model a typical low earth orbit satellite, the Canadian Earth-observation satellite RADARSAT-2. The testbed captures orbital and attitude dynamics, including J_2 gravitational, drag and solar radiation pressure perturbation forces and gravity gradient, drag, solar radiation pressure and magnetic torques. The modelled satellite has a GPS for position sensing and a gyro, magnetometer, two Sun sensors and two star trackers for attitude sensing. Its attitude actuators are reaction wheels and magnetorquers. A thruster model has been integrated. The testbed satellite uses Extended Kalman Filters for position and attitude state estimation. A nonlinear proportional-derivative controller is used to control the reaction wheels. The magnetorquers are controlled by the B-dot algorithm for de-tumbling. Tests performed with the testbed showed that it achieved comparable position and attitude determination accuracy to RADARSAT-2. Attitude control accuracy was not equal to RADARSAT-2 but was close enough to suggest a comparable accuracy could be attained with more tuning or the addition of an integral term. Magnetorquer detumbling appears to work. An orbital maneuver planner was implemented with a linear programming formulation. The testbed was applied to some preliminary autonomous collision avoidance problems to determine fuel consumption as a function of warning times prior to a collision. An algorithm was also presented based on in-situ detection of very small debris that future work will be based on. The validation for attitude control under small and large disturbances for nominal and detumbling operations were performed. The planner for orbital maneuvers also appears to work as desired. Then, the testbed was applied to the problem of determining how far in advance an avoidance maneuver should be realized and the consequent impact on fuel consumption. Future work will have the testbed address autonomous avoidance maneuvers given that most of the algorithmic components are developed. The testbed can be used to assess a variety of satellite autonomy algorithms, in addition to autonomous satellite avoidance, with the intent to assess the value of autonomy that is distributed between the satellite and the ground control station.
Date of Conference: September 11-14, 2018
Track: Astrodynamics