Speaker
Prof.
Marcello Romano
(Naval Postgraduate School)
Description
An experimental campaign has been conducted to evaluate the performance of close proximity and docking maneuvers of two different guidance and control algorithms based on Model Predictive Control and on Inverse Dynamics Control. The metrics of performance includes fuel usage, time to target, computational burden and constrain handling.
The experiments have been conducted on two ~10 kg Spacecraft Simulators that float via air-pads over a 4-by-4 meter polished granite monolith surface recreating a reduced gravity and a quasi-friction-less motion in two translational and one rotational degrees-of-freedom (planar motion). By using eight cold-gas thrusters and a reaction wheel, the Spacecraft Simulators are capable of autonomous motion over the floating surface. An onboard tank of compressed air (propellant), a power system and on-board computer give full autonomy to the Spacecraft Simulators. All the required processing (sensor readings, communications, navigation, guidance and control, and actuator commanding) is handled on-board in real-time. The experimental set-up will be described in detail. The navigation problem has been considered solved with the Spacecraft Simulators sensing their position and attitude by using an overhead optical positioning system (VICON) augmented by an on-board Fiber Optics Gyroscope. The accuracy of the Spacecraft Simulator position and attitude knowledge (as well as the knowledge of the target state) can be artificially deteriorated to simulate real sensor limitations and constraints (e.g precision and field-of-view of a RADAR system).
With the Model Predictive Control framework, a cost function (e.g fuel consumption) subject to system dynamics and constraints (e.g. maximum available control actuation level and obstacle avoidance) is minimized over a discretized time period with a finite prediction horizon. By solving the optimization problem, a control input for each discrete time sequence is generated.
In the Inverse Dynamics Control approach the trajectory of the spacecraft is simplified to a known function depending on a set of parameters (e.g. a polynomial). The desired initial and final conditions are then imposed on that trajectory and the parameters that are left unset are then optimized to meet other constrains whilst minimizing a cost function (e.g. fuel consumption). The control input to follow the prescribed trajectory is then generated from the prescribed trajectory and applied to the system.
Both guidance and control approaches eventually reduce to two different non-linear optimization problems that need to be solved periodically in order to provide the control inputs. In this study, the open-source Interior Point OPTimizer (IPOPT) software package has been used.
The set of test scenarios that have been conducted are designed to represent a wide set of rendezvous and proximity operations scenarios (unconstrained and constrained, cooperative and uncooperative docking and proximity operations with or without obstacle avoidance). The goal has been to define a set of tests that can be used to benchmark different guidance algorithms so that a meaningful comparison of different approaches can be made.
| Applicant type | First author |
|---|
Primary author
Prof.
Marcello Romano
(Naval Postgraduate School)
Co-authors
Dr
Hyeongjun Park
(Naval Postgraduate School)
Dr
Josep Virgili-Llop
(Naval Postgraduate School)
