Ussally enhancing an existing technology that is widely deployed and understood to fit a new situation is better than inventing something wholly new (though not always)
Celestial navigation works on the Earth's surface (or the Moon's surface), because being able to determine the orientation of the local horizon (or zenith) is equivalent to determining your latitude and longitude. But that doesn't work for a spacecraft that doesn't have a horizon reference.
Of course, if you're orbiting the moon and you can accurately observe the directions to landmarks such as mountains and craters, you can fix your position relative to them. But that's not really what you'd call "celestial navigation".
I assume you would need orbital ephemeris info much like we need for GPS satellites.
But, given measurements of angles to multiple planets, how well could you estimate your position? Would there a lot of error for your normal vector to the earth's ecliptic plane?
> Sextants can be read accurately to within 0.1 arcminutes, so the observer's position can be determined within (theoretically) 0.1 nautical miles (185.2 meters, or about 203 yards). Most ocean navigators, measuring from a moving platform under fair conditions, can achieve a practical accuracy of approximately 1.5 nautical miles (2.8 km)
Some napkin math, assuming using a Sextant to achieve similar accuracy of 0.1 arcminutes on the Moon, because Moon is about 3.7 times smaller than Earth, that 0.1 arcminutes is around 50 meters on the Moon. One can expect extremely clear sky and certainly not riding waves on the Moon, so the practical accuracy should be close.
[0] https://theaviationgeekclub.com/the-sr-71-astroinertial-navi...