If you look at the series of images DART took as it approached Dimorphos, the asteroid jumps around a bit as you approach. Does this mean small corrections were being made right up till near the end? Assuming that's correct, why is my intuition wrong that you should be able to get it perfectly lined up from further away than that, considering the predictability of the space environment?

  • 5
    $\begingroup$ No specific knowledge, but it's probably deadbands in the control system. $\endgroup$ Sep 27, 2022 at 22:17
  • $\begingroup$ Someone supposedly on the DART team says it was an image processing artifact: "It was not control system instability; it was image processing auto-cropping." $\endgroup$ Sep 29, 2022 at 20:04
  • $\begingroup$ Per Dr. Richardson (working group lead for DART): "with two minutes left before impact, all the thrusters turned off and it was free fall straight into our target asteroid". $\endgroup$
    – BowlOfRed
    Sep 30, 2022 at 6:12

2 Answers 2


This is an extended comment to your self-answer, so I'm making it community.

From your answer,

As to why they weren't able to do a single, early course correction and just glide smoothly in for the last hour, I think perhaps I just expected too much.

I think perhaps you not only expected too much, but that you expected far too much. I would not be at all surprised if the ephemeris for the asteroid was off by kilometers. DART did not go into orbit about the asteroid. It instead went straight for the kill.

From your question,

Assuming that's correct, why is my intuition wrong that you should be able to get it perfectly lined up from further away than that, considering the predictability of the space environment?

Your intuition is wrong because you must have assumed incredible accuracy for where the spacecraft thought it was and for where the spacecraft thought the asteroid was. I would not be at all surprised if the ephemeris for the asteroid was off by kilometers. I would be very surprised if the ephemeris was off by hundreds of meters or less. The spacecraft hit a target about 1/6 of a kilometer across, and from the video, it hit it dead center.

Finally, regarding why you aren't able to find much detail regarding how the DART guidance, navigation, and control works, the answer is on the Applied Physics Lab's description of DART,

In many ways, it was like developing a self-guided missile — a process APL has a rich history of doing.

This is perhaps the quintessential example of dual use technology. The technologies used by DART are indeed very similar to the technologies used in a self-guided missile. If not classified, the technologies used by DART are definitely subject to International Traffic in Arms Regulations. The descriptions of DART are intentionally dumbed-down.

  • $\begingroup$ Even though there are certainly some parallels between DART and missiles, there are also huge differences. The navigations systems work very different, targets are different, air resistance, thruster dynamics... I doubt adapting the DART algorithms for a guided missile would require less work than designing them from scratch. Perhaps this dumbing down has more to do with officials not wanting to give the impression that NASA is publishing away information that could be used by terrorists, than this actually being something to worry about. $\endgroup$ Sep 29, 2022 at 15:12
  • 1
    $\begingroup$ @leftaroundabout I disagree. While the control gains may be very different, the algorithms for making control decisions and algorithms that find those control gains may well be the same. While the navigation algorithms may appear to be very different, that's appearances only. Guided missiles and asteroid-smacking spacecraft both use Kalman filters tuned using very similar algorithms. The same goes for guidance algorithms. The annual GN&Ski conference conveniently held at a nice ski resort inevitably holds classified sessions where the ski gloves, ITAR gloves and classification gloves come off. $\endgroup$ Sep 29, 2022 at 15:24
  • 1
    $\begingroup$ The article clearly says that no course corrections were done for the last 500 miles to avoid blurring the images, and that video is from closer than 500 miles. So it does not actually answer the question. $\endgroup$
    – Jan Hudec
    Sep 29, 2022 at 18:53

I'm revising my answer based on the comments and on the community answer started by David Herman.

Corrections were not made right up until the end, but stopped with 500 miles to go. This is according to the Applied Physics Lab's description, cited in the community answer. According to this NASA site, the last frame where you could see all of Didymos was from a distance of about 570 miles. So, by the time Didymos was no longer visible, corrective thrusts (and this includes changes to orientation as well as translation) should have stopped.

I think the most plausible explanation is the one given by a Reddit commenter that it was inconsistent auto-cropping. The link I gave isn't an authoritative source, but the whole reason they stopped thrusting was to provide a stable imaging platform, so it seems likely they would have gone to some effort to prevent some kind of camera oscillation from continuing.

I thought for a while that "wobbling" would be impossible because the craft is subject to conservation of angular momentum. But the solar panel arrays are quite long, so the camera could wobble relative to them. But the first article I mentioned says that they worked hard to minimize that. So I think it's just cropping.

The Reddit link has a nice video, by the way. And the first article mentioned is worth reading. It has a simulated video that actually has corrections continuing until well after the simulated Didymos is out of the picture. So a possible alternative hypothesis is that thrusting was actually occurring for longer than my previous analysis suggests. I don't have a theory on how that could be, except for my general lack of understanding.

As to why they couldn't do one clean, early burn, I think the community answer covers that well. Fundamentally they never knew the relative positions well enough to make that calculation.

  • $\begingroup$ As far as I understand, the ion engine was partially a tech demonstrator. When they used it, they noticed some adverse effects on some other electronic component (I forgot what it was exactly, it might have been a power fluctuation or some heating). It turns out that the Falcon 9 performed so precisely that they didn't actually need the ion engine. $\endgroup$ Sep 28, 2022 at 9:25
  • 5
    $\begingroup$ I doubt that these are actual course corrections - that would have been too much dv. I think they rather are corrections of the orientation of the spacecraft to keep the camera pointed correctly. $\endgroup$
    – asdfex
    Sep 28, 2022 at 9:30
  • $\begingroup$ @asdfex From the APL web page of DART (see my community answer), they most definitely were both attitude and translation corrections. $\endgroup$ Sep 28, 2022 at 13:12
  • 1
    $\begingroup$ @DavidHammen There might be some translation corrections as well - but they wouldn't cause the jitter the question is about. Especially not if the claim of 1/10000 g is correct. $\endgroup$
    – asdfex
    Sep 28, 2022 at 13:19
  • 1
    $\begingroup$ The article linked in the other answer actually says course corrections were stopped 500 miles from the target to avoid blurring the final images. So orientation corrections or wobbling of the spacecraft. $\endgroup$
    – Jan Hudec
    Sep 29, 2022 at 18:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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