Unscented Trajectory Optimization

What is it about?

Unscented trajectory optimization is a fast new technique that helps mitigate the risks associated with the use of deterministic methods. It can be particularly useful to recover failed missions when the failures are caused by sensors and/or feedback loops. This paper uses a ship navigation problem (known as a Zermelo problem) to illustrate the concepts. When these ideas are applied for an attitude guidance problem involving NASA's Hubble Space Telescope, it reveals that the mission can be fully recovered under failure of all gyros. Together with sensitivity analysis, unscented trajectory optimization can be used for initial mission design to provide decision makers quick estimates of risk, reliability and confidence levels so that ``optimal missions'' do not suffer from cost overruns due to requirements creep. In addition, the operational implementation of unscented controls can be used to enhance the safety of autonomous operations at distant celestial bodies particularly if/when there are failures in sensors or inner-loop feedback controls.

Featured Image