Pulsars can be used for autonomous spacecraft navigation by obtaining position and direction in space with an accuracy of 2km, on condition the spacecraft is provided by one small X-ray telescope on board.

"New method operates autonomously, increasing the number and capabilities of space missions", according to a news page from University of Leicester.

Fascinating stuff; this would be an upgrade from star trackers, which can get confused rather easily by nearby debris as we saw with Rosetta!
Then the question becomes one of mass; is a small telescope lighter/equal to a bunch of star trackers?

The title first made me think of something by DLR's Advanced Study Group that proposed creating a "solar-system positioning system" using beacons to be placed somewhere around the Kuiper Cliff... within the next 150 years.

As for mass, star trackers run around 1.0-1.2 kg per unit. There are designs for ultra-low-mass x-ray telescopes for picosatellites that are in the same weight range.

Just for comparison, how accurate are the usual Sun + Canopus methods? I know that Cassini is equipped with an inertial measurement platform, so I assume that star locks are used to update that much like GPS-INS combos on aircraft.

These only sense orientation, not position.

Deep Space 1 used visible asteroid imaging to autonomously determine its position.

Yeah, realized that was dumb as soon as I typed it, but it was too late to go back at that point; thanks!

Re orientation and position: I'd guess, that x-ray telesecopes will tend to have a narrow fov, and few valid x-ray sources (?). So, in order to be able to observe the pulsed x-ray sources for obtaining position data, they'll likely still need additional star trackers to recover orientation.

There's actually an experiment launching for the ISS next year that will test X ray navigation, SEXTANT - https://en.wikipedia.org/wiki/X-ray_pulsar-based_navigation

I still don't quite understand how it will work, if pulsars just emit identical pulses.

Pulsars are like very precise clocks. They emit short signal peaks. Each pulsar has its own period. This is like a lighthouse. So you can identify the pulsar by its signature. Combine the highly predictable timing of the pulses with light travel times to infer the distance from e.g. Earth, or better from some system closer to inertial like J2000, by measuring the signal phase shift. Another way to measure the signal phase shift is by comparing the "pulsar times" with an on-board atomic clock; so you get independent from a reference signal from Earth.
Combine the distance data to calculate your 3d-position.
This is essentially the same principle like that of GPS, just a little simpler, since you don't need to include rapidly changing relative motion of the senders into your considerations.

The Doppler shift, i.e. changes of the pulse frequency, helps to determine your relative velocity, and hence your location in a second, although related, way.

Thanks, I find this really interesting.

So the signal phases form a kind of grid in space on the order of the wavelengths? Which for a millisecond pulsar would be ~c/f = 3e8m/s / 1000Hz = 3e5m = 300km.

So I guess they could determine position at a higher resolution depending on how accurately they could measure the phase, and more pulsars would give more gridlines and so more accuracy (if I'm understanding correctly).

Well, the pulses aren't sinus-like, but very short and intense. This results in an accurate time resolution. The time between the peaks can be interpolated by a clock.
More pulsars tend to reduce statistical errors or outliers, yes.