Autonomous navigation started in early phases of deep space exploration as a provision to achieve accuracy for planet/natural satellite targeting and approach, where radio tracking accuracies were low, as well as ephemeris data was not quite up to accuracy marks. AutoNav and OpNav were the initial systems used by DS1, New Millenium and Cassini among others.

These autonomous techniques employed a camera which would scan primary body w.r.t an inertial background of fixed stars. Basically, to find the position of the spacecraft. Since then, various autonomous systems have been developed.

Now, for orbit determination, total state (velocity and position vector in a reference frame) information is a prime necessity.

So, as far as I can tell, each of autonomous techniques, if employed for a deep space mission would require the support of Deep Space Networks for velocity determination. Is it so? Are there any other provisions(Methods, Geometrical or others) to determine velocity or any component of velocity autonomously on-board?

  • $\begingroup$ Are you sure you have checked this question: space.stackexchange.com/questions/942/… $\endgroup$ Aug 20, 2015 at 22:46
  • $\begingroup$ By differentiating your position you get the velocity vector. $\endgroup$ Aug 20, 2015 at 22:47
  • $\begingroup$ Doppler shift measurements from dozens of quasars can also help a bit. $\endgroup$ Aug 20, 2015 at 22:51

1 Answer 1


Positions over time can be used to determine velocity, not just in the obvious way but more accurately than that using a model of the gravity field the spacecraft is moving in, dominated by the Sun for deep space vehicles.

It is possible to develop a system to navigate a deep space vehicle fully autonomously, using only a camera for data and designing its own maneuvers or its own ion thrust profiles on-board to achieve an objective.

I figure we will see common use of this capability when we have hundreds of deep space vehicles out there. When we have something closer to a dozen, like we do now, it is not onerous to use (super accurate) radiometric tracking and design maneuvers on the ground. So that's what we do now, which gives greater peace of mind.

  • 2
    $\begingroup$ "Positions over time can be used to determine velocity, not just in the obvious way but more accurately than that using a ... " If i could know more about it because i'm guessing it's not just differential position measurements that are used for velocity determination. Any papers/references? $\endgroup$ Aug 21, 2015 at 8:43
  • 2
    $\begingroup$ For a start, see Lambert's Problem. In the simple case of motion in the gravity field of a single object (e.g. the Sun), given two positions and the time between them, you can determine the orbit. More generally see Orbit Determination. When doing this sort of thing, one must keep very careful track of how the uncertainties in the observation propagate into uncertainties in the solution. $\endgroup$
    – Mark Adler
    Aug 21, 2015 at 15:23
  • $\begingroup$ Talking about Lambert's Problem, we see that it restricts "trajectory/orbit dynamics " to two-body problem. What if we were exploiting three body or four-body dynamics for a trajectory design? $\endgroup$ Aug 22, 2015 at 21:54
  • 1
    $\begingroup$ Then you fit the trajectory numerically with all of the gravity fields. You might use Lambert's solution for the dominant body as the initial guess. $\endgroup$
    – Mark Adler
    Aug 22, 2015 at 22:06

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.