An Autonomy Design Competition for Non-Cooperative Space Operations in Kerbal Space Program

Sign Up Now

Join us for the third annual software design competition where participants develop autonomous agents for maneuvering satellites engaged in non-cooperative space operations.

  • The software design challenge is based upon MIT Lincoln Laboratory’s Kerbal Space Program Differential Games (KSPDG) open-source library.
  • Participants in the challenge will develop agents and algorithms for autonomous control of spacecraft simulated using Kerbal Space Program.
  • The agents will seek to solve several challenging problems related to non-cooperative space operations, such as pursuing an evasive satellite and conducting multi-satellite proximity operations.
  • Participants are encouraged to develop agents using a range of technologies, such as reinforcement learning, optimal control, and game-theoretic techniques.
  • Participants’ agents will be evaluated on a range of metrics, including mission completion time, fuel consumption, and relative distances achieved.
  • The participant pool will undergo a series of elimination events and teams with the best-performing agents will be invited to present their agents via a live demonstration during AIAA SciTech Forum in January 2026.
  • Monthly meetings will be held in the lead up to AIAA SciTech Forum to allow technical exchanges between participants, MIT Lincoln Laboratory developers, and AIAA SciTech Forum organizers.

There is no cost to participate, and neither AIAA membership nor AIAA SciTech Forum registration is required.

About the Competition

As the orbital domain becomes more congested, competitive, and contested, the ability to rendezvous with and service malfunctioning satellites is essential to maintaining a safe space environment. Servicing operations become dramatically more difficult if the malfunctioning satellite is equipped with autonomous collision-avoidance capabilities. Such a scenario does not lend itself to solutions from existing orbital mechanics or optimal control theory; instead, it must be treated in a game-theoretic fashion, where each satellite is considered a distinct player in a game of cat-and-mouse.

What maneuvering strategies by the servicing (pursuing) spacecraft ensure a safe rendezvous, and what maneuvers by the malfunctioning satellite (evader) would be most challenging to overcome? Such questions are best answered in the context of game theory—more specifically, the field of differential games.

MIT Lincoln Laboratory has recently released a set of open-source differential game challenge problems for the orbital domain. Using the Kerbal Space Program (KSP) game engine, a set of pursuit-evasion benchmark scenarios has been developed that will require sophisticated AI and autonomous control algorithms to solve.

The purpose of this session is to host a software design competition that will inspire the development of autonomous control techniques necessary to solve such orbital differential games and then showcase the best techniques in a final round of competition held during the session.

Competition participants will submit their software agents for real-time evaluation during the session, and winners will be announced at the end. Other session attendees will be able to observe the final round of the competition and the tactics of various participant agents via a projection of the KSP competition environment.

Questions? Contact Alex Mercer.

captured-satellite-nasa

AIAA Corporate Partners