TL; DR

A system for calculating optimal collision avoidance maneuvers for satellites based on reinforcement learning.

About

We made Space Navigator - a system for automatically determining the probability of collision of space objects and calculating the most effective evasive maneuver using AI.

Stage of the project:

• Research Article “Space Navigator: a Tool for the Optimization of Collision Avoidance Maneuvers” (2020)
• MVP (2019)
• First place in the competition of projects of the Academy of Roscosmos (presentation slides, 2018)
• VR presentation (2018)

Demonstrations and reports:

• “SMILES”, summer school (Moscow, Skoltech, 2020)
• “Space debris: fundamental and practical aspects of the threat”, Russian conference with international participation (Moscow, Institute of Astronomy RAS and Space Research Institute RAS, 2019)
• “IAA SciTech Forum”, international conference (Moscow, RUDN, 2018)

Mentions:

• Yandex Academy article (2019)
• Известия (news) (2018)

Details can be found in the scientific article.

My role

• Management
• Development: ML, RL, Web API
• First author in the article “Space Navigator: a Tool for the Optimization of Collision Avoidance Maneuvers” (2020)
• Most of the code and experiments
• Сonference speeches