ESA GNC Conference Papers Repository

Optimal real-time force rendering for on-orbit structures assembly
C.A.P. Pascucci, M.S. Szmuk, B.A. Acikmese
Presented at:
Salzburg 2017
Full paper:

Autonomous rendezvous, docking, and proximity operations are key capabilities that will be designed into the next generation of spacecrafts. These capabilities enable missions such as the refueling and servicing of on-station assets, or the assembly of structures too large and heavy to launch on a single launch vehicle (e.g. like large aperture space telescopes), where several modules are launched into orbit separately and then precisely linked together. The benefits of this approach range from reducing human-generated space debris, to enabling otherwise insurmountable missions. These future space systems will require a new level of autonomy to reduce mission planning efforts, increase mission flexibility, and tolerate greater uncertainty inherent to proximity operations. Specifically, this will require a new class of GN&C systems that can satisfy mission requirements while guaranteeing that vehicle operational constraints are not violated. Over the past decade, convex optimization has been shown as a viable approach for path planning of dynamical space systems, and for aiding in high level mission planning decision making. Moreover, recent advances in both hardware (e.g. with low-power, high-performance embedded CPUs) and software (e.g. with new effective problem solver implementations) has reduced the execution time of convex optimization algorithms to between millisecond and a few hundreds of milliseconds range. This will enable the real-time implementation of such algorithms, allowing to use convex optimization not only as a planning tool in the design phase, but also on-line during flight. In this paper we focus on the proximity operations and docking, where to avoid plume impingement on the target, direct backfire braking must be avoided. To do so, we propose to synthesize the desired braking force without firing the chaser's thrusters towards the target. This will allow for higher initial approach speeds, hence increasing the feasible rendezvous envelope, while preserving operational safety. By formulating the problem in the Second-Order Cone Programming (SOCP) framework, we can generate fuel-optimal trajectories that lie in a specified approach cone, and that satisfy attitude constraints that prevent undesired plume impingement on the target. Furthermore, our problem formulation can target a final-relative-speed range that is sufficiently low to avoid damage, but sufficiently high to activate latching mechanisms, and can account for control limitations such as minimum and maximum engine thrusts. The performance of the proposed algorithm will be demonstrated with simulations in Low Earth Orbit (LEO). Moreover, to assess the real-time capabilities of our algorithm, we will measure CPU timings for trajectory generation on embedded processors, exploiting a custom SOCP solver developed in previous works by some of the authors.