Space-based assembly with symbolic and continuous planning experts

When considering the necessary role automation must play in future space missions and endeavors, it is important to study basic scenarios that further our understanding of the challenges we must overcome to meet such objectives. We believe some of these challenges lie in the inherent disconnect between the planning of tasks and the development of the continuous trajectories that must be followed to accomplish these tasks. While terrestrial action choices to optimize fuel use and time can be implemented with traditional AI planning tools, the optimization of these resources with respect to space operations requires full-state trajectories -i.e., geometric motion (a path) that includes point velocities/accelerations governed by nonlinear differential equations of motion. The difficulty in this is perhaps no where more apparent than in space-based assembly with self-actuated components that rendezvous/dock, as the action choices for each component and its corresponding trajectories scale with the number of assembly members. By integrating optimal task planning with specialized astrodynamics "experts" for optimal orbit transfer and proximity docking operations, we show that a trajectory planning library with an application to self-assembly can be developed that is both capable and efficient