Spacecraft localization beyond Earth orbit

Question

How does a spacecraft localize itself after leaving earth orbit? In robotics you can either use known broadcasting information or do real-time position tracking(some type of SLAM) but I suspect that probably isn't going to work in this case given the scale.

Answer 1

Spacecraft in between planets oftentimes do not know where they are, and if they do it's because they maintain a rough estimate based on more precise estimates occasionally sent from Earth. The key problem with maintaining position and velocity state is that it is impossible to measure gravity. Gravitational acceleration has to be estimated based on position and velocity. Non gravitational acceleration can be sensed; this is what an accelerometer measures. The estimated gravitational acceleration plus the sensed non-gravitational measurements can be integrated to update velocity estimates, which in turn can be integrated to update position estimates. This is called dead reckoning. Occasional fixes are needed lest the vehicle become dead.

Spacecraft oftentimes do know where they are pointing (attitude). Pointing information can easily be obtained using star trackers, with a once per a tenth of a second to a once per a ten second update rate, and maintained in between star tracker updates using rate gyros. Dead reckoning is a much less significant issue for attitude as opposed to location, for many reasons. First and foremost, the updates can be obtained on the spacecraft. Even low-cost spacecraft are now outfitted with star trackers. Secondly, the updates are frequent. The drift with a ten second lag between star tracker updates is not that much. Thirdly, the integration is direct (sensed rate integrates directly to attitude) as opposed to rather indirect (sensed+estimated acceleration integrates to velocity, which in tern integrates to position).

This is problematic for vehicles that will land on another object, be it a planet, moon, asteroid, or comet. Landing with some landing legs two meters higher than others is a recipe for a tip over disaster. Here, local state information is much more important than global state information. Landers use a number of sensors and algorithms, mostly vision-based, to perform landing. A big problem is avoiding hazardous terrain. The Apollo program relied on human eyes and human brains. Hazard avoidance while landing an autonomous vehicle on a planet, moon, asteroid, or comet -- that's challenging.

Answer 2

Usually the spacecraft position is computed on Earth and therefore it does not know where it is. This computation is made using Deep Space Network (DSN) where we send a signal to the spacecraft which sends it back to us. From this signal we can then calculate its position and send back a maneuver if necessary. The problem is that these signal exchanges can be long and it can quickly become a problem when the spacecraft is far away.

If the spacecraft knew its position, this would eliminate the processing on Earth and the satellite could navigate itself, by sending just one signal from Earth to the spacecraft. However, this requires having a very precise clock on board. GPS satellites carry atomic clocks to help us get to our destinations on Earth, but those clocks require updates several times a day to maintain the necessary level of stability. Deep space missions would require more stable space-based clocks.

The Deep Space Atomick Clock has been developed for this, you can see this article for more information.