Spacecraft Configuration Optimization

A spacecraft can be designed and manufactured more quickly if it is built from a mix-and-match set of subsystems. A spacecraft needs several dozen subsystems, including a propulsion system, an attitude controller, a power bus, a power source, solar panels, flight computers and scientific instruments. NASA's Pluto Express spacecraft project will allow each subsystem to be chosen from an existing supply of pre-designed, pre-manufactured modules, thereby reducing mission costs and increasing mission reliability. The optimization problem is: given a set of mission constraints, choose the best set of modules to implement the subsystems.

This is an interesting combinatorial space because:

• each module decision affects many other module decisions;
• the mission constraints are complex non-linear functions;
• many module decisions may be refined numerically, e.g. the size of a solar panel and bandwidth of a communication system; and
• evaluating a given configuration can be computationally expensive.

It is also an immensely worthwhile problem to solve because of the difficulty and expense of making these decisions manually, and the large cost impact of a successful system.

In conjunction with Steve Chien and researchers at the NASA Jet Propulsion Laboratory, I will formulate this optimization problem as a state-space search and generate evaluation functions that improve search performance. I will also have the opportunity to work on another JPL project, the Integrated Probe Dynamics Tool, which includes a similar component for optimizing the design of a penetrator (a needle-nosed projectile which is sent crashing into a planet to analyze soil).