Skip to content
Marcus Märtens edited this page Mar 11, 2019 · 9 revisions

Multi-objective trajectory optimization for GTOP

The GTOP database (Global trajectory optimization Problem database) was established by the Advanced Concepts Team (ACT) with the goal in mind to provide ongoing challenges for trajectory optimization. Given an easy interface to selected mission concepts, participants could search and submit solutions which were ranked according to their accumulative Delta-V. The GTOP database inspired researchers from Europe and above to improve their algorithms for getting a place in the leaderboard, but has been discontinued as the necessary software has no longer be maintained.

The long term goal of the ACT is to re-establish the GTOP as a modern online portal, which will enable users to publish trajectory problems and solutions online, further increasing collaboration, competition and open exchange.

Pykep (and the related pagmo open source project) are the key enablers for this portal. The goal of this project is to make pykep ready to handle the new GTOP trajectory problems, i.e. functionality that allows a reliable verification and ranking of submitted solution. This includes new methods in multi-objective optimization, to allow ranking trajectories not only by their Delta-V, but also by other performance criteria (i.e. time of flight, maximum thrust impulse) along multiple dimensions. Constructing the set of optimal trade-offs ("Pareto-front") is of essential importance in mission design, as it allows to select fly-by sequences or deep space maneuvers, i.e.:

Märtens, Marcus, and Dario Izzo. "The asynchronous island model and NSGA-II: study of a new migration operator and its performance. " Proceedings of the 15th annual conference on Genetic and evolutionary computation. ACM, 2013.

Consequently, performance evaluation metrics from the field of multi-objective optimization need to be accessible. These may include

Finally, the interface between pykep and pagmo needs to be improved, for example by implementing above performance evaluation metrics in pagmo and transfer new multi-objective algorithms like.

  • SPEA-2
  • Multi-objective PSO

This will not only present a precondition for the establishment of the GTOP portal but will also the design and study of trajectories, for which a careful analysis of trade-offs and constraints is at the heart of any decision making process.

Clone this wiki locally