On their website, SpaceX claims that Starlink satellites are able to conduct autonomous collision avoidance manoeuvres using alerts from the US Department of Defence. This seems very reasonable, given how many Starlink satellites are planned/already up there and given how labour intensive the standard, manual system is (NASA, ESA).

I am struggling to find any information on how their system works and was hoping someone here could point me to some technical papers. (My google scholar search brought up nothing, but I might be using the wrong keywords.) I am also wondering if anyone is aware of any other satellites that are able to (or planning to be able to) do autonomous collision avoidance manoeuvres.

Q1: Are there any technical papers on what techniques SpaceX is using for their collision avoidance systems on Starlink satellites? Or is their research mostly happening behind closed doors?

Q2: Are there any other satellites capable of autonomous collision avoidance?


I don't think there are any papers about it, but here's what I've gleamed from my studies on it.

  1. As you mentioned, they use information from the DOD, specifically Space-Track, or C-SPOC, or J-SPOC. They are all really the same thing...
  2. Space Track will send out updates when predicted close approaches may happen to the satellite operator.
  3. SpaceX somewhere must be receiving this information and determining when to make a maneuver. There's no way this is done on every satellite, it's complex work, and easier done on the ground.
  4. The system will automatically determine how to respond to the close approach notices. I'm not sure if this is via the ground or the satellites are informed of the potential close approaches, in either case some thrusting commands are made to avoid the close approach.
  5. After some time, the satellite will slowly be commanded back to its normal position. This one might happen automatically on the satellite itself.

To my knowledge, no other operator does this so automatically. For every other one, there is at least a human in the loop (And there may even be for Starlink too). But these kinds of things just don't happen often enough to merit automation for most networks. I know that Iridium uses Space Track information, and then has a team that will decide what to do.

Lastly, if a close approach happens with two active satellites of different origins, it is polite to coordinate efforts between the two organizations. I'm sure SpaceX does that. The last thing you would want is both of you to thrust "away" in the same direction and hit.

| improve this answer | |
  • 2
    $\begingroup$ except when they don't teslarati.com/… and spacenews.com/… $\endgroup$ – uhoh Jun 12 at 14:55
  • 2
    $\begingroup$ @uhoh that event suggests that there is a human in the loop and implies that autonomous collision avoidance is a buzzword (at least for the lower probability collisions) $\endgroup$ – aranedain Jun 12 at 15:08
  • 1
    $\begingroup$ @aranedain or just that it wasn't turned on yet... $\endgroup$ – uhoh Jun 12 at 15:11
  • 1
    $\begingroup$ @uhoh ah yes, the ol' software is installed, but not turned on until the check clears gimmick:) jk $\endgroup$ – aranedain Jun 12 at 15:15
  • 1
    $\begingroup$ @uhoh That was with the very first batch of satellites, pretty early on. Pretty sure they didn't fully trust things, and there was more to that story, like conflicting values coming from Space Track. $\endgroup$ – PearsonArtPhoto Jun 12 at 22:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.