In the video clip of this interview, captioned:

General Wayne Monteith, 45th Space Wing Commander, talks about the Autonomous Flight Safety System, or AFSS -- a new, automated way of having a rocket self-destruct. Video by Malcolm Denemark. (emphasis added)

in the Florida Today article Only on Falcon 9: Automated system can terminate SpaceX rocket launches, General Monteith says (my attempt at transcription):

"Autonomous flight safety system is a game-changer. It increases public safety, it increases our throughput or capacity on the range, and reduces cost to the customers, and eventually cost to the air force. [...] we essentially remove the man or the woman out of the loop, which saves us about three and a half seconds of decision time during flight if there is an anomalous profile. So if the flight is going off-course, it gives the rocket an extra three and a half seconds to get back on course, and correct itself."

Later in the video the self-destruct option is also mentioned.

I really have two questions.

  1. Despite the lengthy list of things described in the interview that have to happen when the decision is made on the ground, is any significant fraction of the three-and-a-half second benefit really coming from moving the decision from ground to rocket, or is essentially all of it really coming from moving the decision from human to computer, and the computer could be on the ground or in the rocket without more than a small fraction of a second difference?

  2. The nature of spaceflight is such that the rocket's flight system is in fact making life-and-death decisions continuously. But with a crewed flight (or perhaps sooner than anticipated a flight with passengers), would the Autonomous Flight Safety System or AFSS now be able to initiate a self-destruct that would include the crew?

  • $\begingroup$ I skimmed that article and if It did not mention it, the SpaceX Crewed Dragon has an safety abort system that will separate the crewed capsule from the "rocket" and propel them to safety. Blue Origin was contracted by NASA to develop a Launch Abort System that can be used on various "crewed capsules". NASA tested it and it worked as expected. So I am not sure if your question took that into consideration. $\endgroup$ Mar 12, 2017 at 17:24
  • $\begingroup$ @EnigmaMaitreya that is a very important point, and it was made well in (at)Schlusstein's answer. I am not sure if the Autonomous Flight Safety System, or AFSS is strictly as SpaceX feature or if the term refers to something that could be implemented on other spacecraft as well. I'm also interested in the first part of the question - is the gain due to the distance between the rocket and the location of the decision, or simply the brain or computer that makes it. $\endgroup$
    – uhoh
    Mar 12, 2017 at 17:52
  • $\begingroup$ My Opinion would be it is a within the software making a "caned" decision. My guess is were the software is running is not critical. Assumption if the Capsule sustains damage then we might as well assume a dysfunction of the Escape System. Saying that I see no real advantage for it to NOT be on bard the capsule. $\endgroup$ Mar 12, 2017 at 18:10
  • 2
    $\begingroup$ Manned capsules have escape systems. Why would you not think the first step of a destruct in such a case would be to fire the escape system? Also, remember that the capsule survived the destruct of CRS-7 and would have been recovered had they thought to program the parachutes accordingly. $\endgroup$ Mar 12, 2017 at 21:57
  • $\begingroup$ @LorenPechtel why would you think that I wouldn't? I didn't say either way. According to (at)Hobbes's answer and links therein, as of now AFSS is not yet man-rated, but that is the plan. In addition to taking a possible "crew safely away" signal into consideration, it can also potentially generate a "crew should safely get the heck away right now" signal or not, depending on if the navigation system believes it still has a chance to safely correct the flight path (or other) anomaly. $\endgroup$
    – uhoh
    Mar 13, 2017 at 3:59

2 Answers 2


AFSS was developed for unmanned launches first. A 2004 NASA paper indicates how manned flight would be handled:

Autonomous Flight Safety System (AFSS) is an independent flight safety system designed for small to medium sized expendable launch vehicles launching from or needing range safety protection while overlying relatively remote locations. AFSS replaces the need for a man-in-the-loop to make decisions for flight termination. AFSS could also serve as the prototype for an autonomous manned flight crew escape advisory system.

AFSS isn't man-rated yet (also in Spaceflight Now):

The Autonomous Flight Safety System could be on crewed flights as soon as next year as commercial companies begin launching astronauts to the International Space Station.

The commercial crew missions will be launched under contract to NASA, which will certify the on-board termination computer for piloted flights.

Note that AFSS is not a SpaceX development. This was done at NASA, and it's now available for all launch providers who implement it in their launch vehicles.

  • $\begingroup$ Thanks - this is a much more thorough answer than I'd hoped for. I appreciate the time and effort. $\endgroup$
    – uhoh
    Mar 12, 2017 at 18:56
  • 3
    $\begingroup$ @Joshua I believe there is no way a pilot could take control at that point. If the computer is not able to keep the rocket in the safe zone and on the right trajectory, human pilot won't be capable of fixing it. $\endgroup$
    – jkavalik
    Mar 12, 2017 at 20:25
  • 2
    $\begingroup$ @jkavalik: They planned for that eventuality on the Apollo missions; the failure mode handled was total computer failure. The shuttle theoretically had the ability; however they believed no pilot could do it in early flight. Once in vacuum it was quite controllable. $\endgroup$
    – Joshua
    Mar 12, 2017 at 20:57
  • 3
    $\begingroup$ @Joshua: Keep in mind that computers became a billion times faster, cheaper and smaller in the mean time. (That's pretty literal). That makes excessive redundancy an option. Also, the launch phase where you need this is within the atmosphere and thus low-radiation, which means a lower risk of externally induced errors. $\endgroup$
    – MSalters
    Mar 14, 2017 at 18:44
  • 1
    $\begingroup$ Submitted an edit -- AFSS is by definition a feature built into the launch vehicle rather than a service provided by the range. The Air Force, NASA, SpaceX & (I believe) ULA have all worked to make it a reality, but the implementation is owned by individual launch providers. $\endgroup$
    – Saiboogu
    Feb 28, 2018 at 21:37

No, a computer will not be allowed to blow up astronauts. It may activate a launch abort system for them though, after which a flight termination system might be used depending on the situation.

  • 2
    $\begingroup$ intentionally blow up at least. Someday it may come to pass that a computer system unintentionally blows somebody up. $\endgroup$
    – uhoh
    Mar 12, 2017 at 16:49
  • 6
    $\begingroup$ Sure, but it's not allowed to do that $\endgroup$ Mar 12, 2017 at 16:50
  • 1
    $\begingroup$ @uhoh To add to schlussein's reply I would hope that NASA/the Designers would have understand the system is not an AI as in sentient. So that means, to at least me, that the Auto-Destruct has meet any and all requirements that the "Human" decision makers would make. Implying a less than obvious aspect. IF the Auto-Destruct is computer made, then "Human" controllers are out of the loop and may be very hard to prosecute in a Civil Case. This may be less important to NASA but it may be very important to Launch Companies. $\endgroup$ Mar 12, 2017 at 18:17
  • 2
    $\begingroup$ One key reason for using AFTS on a crewed vehicle is that it improves safety for the crew (and passengers). While the AFTS might intentionally explode the launch vehicle, automatically detecting a very serious problem gives a 3+ second advantage (compared to manual flight termination) to the launch abort system to safely carry humans away from the mess that is about to ensue. $\endgroup$ Mar 17, 2017 at 13:17
  • 1
    $\begingroup$ @DavidHammen these are answer-quality comments. If you have a moment please consider leaving a supplemental answer. While I have a hunch your comments are safer than an average answer, future readers can learn more by seeing these as an upvoted answer. $\endgroup$
    – uhoh
    Mar 17, 2017 at 13:27

Your Answer

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

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