I wonder if it'd make sense these days to put optical navigation beacons in space. This OP solution relies on inertial navigation, because, the star-tracking part is only a reference for orientation in space, which is an incomplete coordinate set. But if you were to put *artificial* stars close to Earth, you'd gain an additional parallax measurement, and obtain a complete basis.
teamonkey · 15h ago
I believe it has inertial systems because it’s designed to be mounted to aircraft carriers, which constantly pitch and roll about, making it harder to get a fix.
If you have an accurate true north compass reading, an accurate clock and assume that you’re at sea level, star positions give you all the information you need to find your location on earth.
> If you have an accurate true north compass reading, an accurate clock and assume that you’re at sea level, star positions give you all the information you need to find your location on earth.
After stopping for a little while, the US Naval College brought back sextant training:
I'm not following your argument; magnetic north and celestial north are redundant. I don't understand what you think a magnetic orientation adds to a star tracker.
The stars look exactly the same from everywhere on Earth. The only thing looking at them can tell you is an absolute rotational orientation—nothing more.
Old-style navigation was based on the differential orientation of stars relative to the local orientation of the Earth's apparent horizon, but that measurement's not accurate enough for modern GNSS. (I don't think).
I don't know how the SR-71 guidance works, but your link describes it as having an inertial component.
Amezarak · 13h ago
I may be betraying my ignorance here but I don’t understand what you mean when you say the stars look the same everywhere on earth. You can definitely determine latitude from what stars are visible in the night sky along with their position in the sky, as, for example, the Southern Cross is only visible in the southern hemisphere; the converse is famously true for the North Star. It makes sense to me you don’t gain anything from having a compass (assuming clear nights) though, since you have the celestial north.
Google claims that human-based celestial navigation done correctly can be accurate to 1 nautical mile, which seems good enough for navigation but not for any weapons targeting. Definitely way worse than GNSS. I would assume there are automated systems out there that can reduce the sources of error together get it down to less than that but probably not several orders of magnitude.
perihelions · 12h ago
> "for example, the Southern Cross is only visible in the southern hemisphere"
It's in the same place everywhere; it's the position of the Earth, which you're implicitly measuring by pointing out that it's obstructing your view, that's different. The things above the horizon are visible; the things below the horizon are not. That's not facetious; the core idea of this type of navigation is to measure the position of the horizon with high precision, which is hard. Very hard if you're surrounded by mountains. ("Which hemisphere am I on?" is... low precision, if easy to measure). Those classic methods aren't pure astrometry: they're differential astrometry that measure the horizon against stars, the relative orientation of the two.
teamonkey · 8h ago
You don’t need to see the horizon at all, you just need to establish a vector that points Up. Traditional instruments used the horizon, but now we have gyroscopes, gravity sensors and satellite surveys of the Earth’s density and magnetic field, none of which require GPS or networking.
You don’t need a wide field of view either, you just need an enough pairs of stars that the plate solver can provide an accurate reading of what in the celestial sphere the camera is actually looking at.
No comments yet
sneak · 15h ago
How would they be powered to be bright enough to be visible?
If you switched to parts of the EM spectrum where you don’t need high power output, you’re back at radio waves and you’ve simply reinvented GPS.
perihelions · 14h ago
Optical isn't particularly power-inefficient, is it? The equivalent of an optical magnitude +5 star, in 500 km low orbit, is (if I didn't err) only 1 kilowatt of isotropic power in a solar spectrum. Much less if it's a non-isotropic beam pointed towards earth.
+5 is the brightest ~1,600 stars, likely similar to whatever any star tracker would be using.
sneak · 5h ago
24kWh daily is a lot of energy.
For reference, the current GPS satellites are 1000kg and transmit with 50W (1.2kWh daily).
Here on the ground, for rough calculations, 50kWh is around 450kg in batteries alone. There might be lighter chemistries but this is just to get you in the ballpark.
I think a 1kW light on 24/7 in orbit would need to be a very large vehicle (2-4x the size of the current Navstar satellites) with huge batteries and gigantic solar arrays. Falcon 9 can probably do it, FH definitely can.
An RTG might work but those are out of fashion near Earth these days for obvious reasons.
In any case it would be quite costly (you’d need lots) and its resistance to jamming would benefit your adversaries just as much as it would benefit you. Not to mention, the astronomers would hate you.
wongarsu · 15h ago
Really bright geostationary satellites to reflect sunlight might work. At geostationary they should be far enough out that you always have line of sight to one that isn't in earth's shadow. You'd need to specifically design them to have a huge surface area to get enough brightness. Maybe that'd make them impractical
Vox_Leone · 14h ago
I'm working on an open-source system called SpinStep (just search GitHub) — it's a quaternion-driven traversal framework for orientation-based logic and spatial data structures.
It’s not directly tied to geolocation, but it could integrate nicely with something like Astradia. Since Astradia provides high-fidelity attitude data without relying on GNSS, SpinStep could use that orientation stream to drive autonomous behavior trees, scanning patterns, or state transitions — all without depending on coordinates or maps. Basically: orientation in, logic out.
Would love to hear from others thinking about orientation-first autonomy or mapless navigation.
That's strange, I submitted the english version but I see it has been changed to the french.
voxadam · 15h ago
It's just a guess, but I'm willing to bet HN mangled the URL you submitted and set it to the URL specified by the hreflang="x-default" header. Now that I think about it, I'm pretty sure I've had the exact same thing happen in the past on a story I submitted.
consumer451 · 12h ago
I am pretty that sure HN automatically reads the canonical tag on a page, and changes the link to that. In this case, that was:
It does say that higher in the atmosphere will always be better from an interference perspective (makes sense, as then there will be less atmosphere between the sensor and the stars), so that is why they market this specifically for use on aircraft. At 250k per unit it's not something you'll be able to mount on your car anytime soon.
rich_sasha · 15h ago
Perhaps it uses just the one star, the Sun, at daytime+clouds. You can get a surprising amount of precision just from that (but not the stated 1m per 70km travel, surely). And it's easy to spot, maybe even under heavy cloud cover.
On clear sky, you can often see Venus and Sirius with naked eye, so surely easier still with a precise instrument.
sorenjan · 15h ago
> not the stated 1m per 70km travel, surely
I don't think that's what they claim. They write "tracking capacity to within a few arc-seconds, equivalent to 1 meter at a distance of 70 km.", so I interpret that as a way of visualizing how small of an angle a few arc-seconds is.
This provides an absolute navigation reference, so a relative error after a certain distance traveled doesn't make sense either. The final navigation system would use inertial navigation combined with this.
throw0101b · 14h ago
> Perhaps it uses just the one star, the Sun, at daytime+clouds.
For marine celestial navigation, there are fifty-eight stars that are regularly used / have been standardized on:
With electronics it may be possible to see more than with the naked eye ("Mark One Eyeball"), and have a larger 'almanac' with digital storage than is practical with paper.
As a side note: using only one star and the Sun would give you two lines of position, and they would potentially intersect at two locations, so if you don't have any estimate of your position you don't necessarily know where you are between the two:
If you do have some idea, then you'd pick the intersection which is closest.
stogot · 15h ago
The e day part surprised me. I was familiar with other systems that use stars in darkness (very high altitude)
hollerith · 13h ago
This could help North Korea and Iran develop ICBMs.
Havoc · 12h ago
US had this tech 50+ years ago (sr71) so would think it’s within reach of most state actors now even without help
aredox · 14h ago
Is there a similar system but much more affordable? I don't need it to be space-ready or milspec-grade or even compact, I was just thinking of something for tinkerers.
(The translation used by Sodern, "Star Tracker", gets me only camera gimbals to take astrophotography pictures)
Edit: found some leads around the cubesat community. Still to see if it works from the ground or in daytime.
And even if your ground transmitters can be taken out, it's a lot easier to build some new Loran transmission gear and radio towers than it is to launch new GNSS satellites (which are also vulnerable to attack).
relaxing · 1h ago
I realize that probably sounds insurmountable, but the practical upshot of that factoid is you can jam GPS with a handheld transmitter. Military jammers can be much, much bigger. As big as a 5,000,000x Loran transmitter even!
I think in reality, it’s easier to put little star trackers on your boats and planes.
And it’s way easier to launch another cruise missile at the rebuilt LORAN site.
contingencies · 15h ago
Use case = ?
3kg! That's a very high price to pay for reliability under GNSS denied environments in the day time. Surely you can get ~altitude from a barometer, ~position from inertial + IMU + prior position, and if you want more accuracy maybe DEM models + topography is better.
Therefore I'd suggest this is not going to be very useful on anything with a view toward efficiency (drones). This is more likely a 'quick fix' plugin for existing fat-plane avionics for 'identified risk' reasons. OK move commercially, but technically meh. The go to market plan would presumably be FUD around GNSS denied situations. Reality: unlikely to be problematic in most cases (known flight path, high altitude, low GNSS-denied environment dwell time).
throw0101d · 14h ago
> Use case = ?
If you tend to operate near / around the yellow and red zones shown here:
That's the annual salary of of 2-3 airline pilots, which you pay year after year in OpEx. This is a one-time CapEx that will work for years and may allow you to fly (i.e., generate revenue) when otherwise you would be grounded:
In a scenario where you have a conflict with completely denied GNSS (e.g. mass destruction of satellites) the applications are basically limitless. 3kg is perfectly viable on anything which isn't a small drone. Aircraft, ships, tanks, APCs, etc. all can use them.
contingencies · 14h ago
Good use case. Although, they said the design is for planes. This is significant because the optics probably assume uninterrupted sky view with no interceding light sources which is an invalid assumption at ground level. If setting off a flare or cycling street lights is all it takes to deny positioning it's a weak terrestrial solution.
relaxing · 12h ago
So just move away from LOS sources of interference.
It is exactly what I mean. That "early" (cold war era and maybe later) spy planes, strategic bombers, cruise missiles and even ICBMs used stars to navigate.
And, oh, yes, all ships before last century, too, of course. So it is more spiral than circle :)
defrost · 15h ago
and post WWII landrovers, the ships of the desert, pre GPS, used star fixes to layout grid roads for the world's largest test range*.
If you have an accurate true north compass reading, an accurate clock and assume that you’re at sea level, star positions give you all the information you need to find your location on earth.
The SR71 used a similar system.
https://theaviationgeekclub.com/the-sr-71-blackbird-astro-na...
After stopping for a little while, the US Naval College brought back sextant training:
* https://www.npr.org/2016/02/22/467210492/u-s-navy-brings-bac...
The stars look exactly the same from everywhere on Earth. The only thing looking at them can tell you is an absolute rotational orientation—nothing more.
Old-style navigation was based on the differential orientation of stars relative to the local orientation of the Earth's apparent horizon, but that measurement's not accurate enough for modern GNSS. (I don't think).
I don't know how the SR-71 guidance works, but your link describes it as having an inertial component.
Google claims that human-based celestial navigation done correctly can be accurate to 1 nautical mile, which seems good enough for navigation but not for any weapons targeting. Definitely way worse than GNSS. I would assume there are automated systems out there that can reduce the sources of error together get it down to less than that but probably not several orders of magnitude.
It's in the same place everywhere; it's the position of the Earth, which you're implicitly measuring by pointing out that it's obstructing your view, that's different. The things above the horizon are visible; the things below the horizon are not. That's not facetious; the core idea of this type of navigation is to measure the position of the horizon with high precision, which is hard. Very hard if you're surrounded by mountains. ("Which hemisphere am I on?" is... low precision, if easy to measure). Those classic methods aren't pure astrometry: they're differential astrometry that measure the horizon against stars, the relative orientation of the two.
You don’t need a wide field of view either, you just need an enough pairs of stars that the plate solver can provide an accurate reading of what in the celestial sphere the camera is actually looking at.
No comments yet
If you switched to parts of the EM spectrum where you don’t need high power output, you’re back at radio waves and you’ve simply reinvented GPS.
+5 is the brightest ~1,600 stars, likely similar to whatever any star tracker would be using.
For reference, the current GPS satellites are 1000kg and transmit with 50W (1.2kWh daily).
Here on the ground, for rough calculations, 50kWh is around 450kg in batteries alone. There might be lighter chemistries but this is just to get you in the ballpark.
I think a 1kW light on 24/7 in orbit would need to be a very large vehicle (2-4x the size of the current Navstar satellites) with huge batteries and gigantic solar arrays. Falcon 9 can probably do it, FH definitely can.
An RTG might work but those are out of fashion near Earth these days for obvious reasons.
In any case it would be quite costly (you’d need lots) and its resistance to jamming would benefit your adversaries just as much as it would benefit you. Not to mention, the astronomers would hate you.
It’s not directly tied to geolocation, but it could integrate nicely with something like Astradia. Since Astradia provides high-fidelity attitude data without relying on GNSS, SpinStep could use that orientation stream to drive autonomous behavior trees, scanning patterns, or state transitions — all without depending on coordinates or maps. Basically: orientation in, logic out.
Would love to hear from others thinking about orientation-first autonomy or mapless navigation.
* https://news.ycombinator.com/item?id=42767797
It does say that higher in the atmosphere will always be better from an interference perspective (makes sense, as then there will be less atmosphere between the sensor and the stars), so that is why they market this specifically for use on aircraft. At 250k per unit it's not something you'll be able to mount on your car anytime soon.
On clear sky, you can often see Venus and Sirius with naked eye, so surely easier still with a precise instrument.
I don't think that's what they claim. They write "tracking capacity to within a few arc-seconds, equivalent to 1 meter at a distance of 70 km.", so I interpret that as a way of visualizing how small of an angle a few arc-seconds is.
This provides an absolute navigation reference, so a relative error after a certain distance traveled doesn't make sense either. The final navigation system would use inertial navigation combined with this.
For marine celestial navigation, there are fifty-eight stars that are regularly used / have been standardized on:
* https://en.wikipedia.org/wiki/List_of_stars_for_navigation
With electronics it may be possible to see more than with the naked eye ("Mark One Eyeball"), and have a larger 'almanac' with digital storage than is practical with paper.
As a side note: using only one star and the Sun would give you two lines of position, and they would potentially intersect at two locations, so if you don't have any estimate of your position you don't necessarily know where you are between the two:
* https://en.jeandusud.com/two-equations-for-celestial-navigat...
If you do have some idea, then you'd pick the intersection which is closest.
(The translation used by Sodern, "Star Tracker", gets me only camera gimbals to take astrophotography pictures)
Edit: found some leads around the cubesat community. Still to see if it works from the ground or in daytime.
https://ieeexplore.ieee.org/document/9179736
https://openstartracker.org/
* Part 1: https://www.youtube.com/watch?v=nkvN74wuT8w
* https://www.youtube.com/playlist?list=PL-_93BVApb5_Gufx3xFN9...
* https://www.glennsmuseum.com/items/b52_astro/
* https://en.wikipedia.org/wiki/Loran-C
China built out a GNSS backup with it:
* https://www.gpsworld.com/china-completes-national-eloran-net...
* https://rntfnd.org/2024/10/03/china-completes-national-elora...
* https://www.mdpi.com/2076-3417/13/23/12703
And there's been some rumblings from Korea and UK:
* https://www.gpsworld.com/south-korea-partners-with-broadcast...
* https://www.gpsworld.com/uk-leading-the-west-in-pnt-with-clo...
but no major moves in most countries, even though there's a recognition of GPS/GNSS vulnerabilities by even the (US) military:
* https://fedtechmagazine.com/article/2022/05/dod-transportati...
https://www.nab.org/bps/
* https://www.militaryaerospace.com/rf-analog/article/14181490...
And even if your ground transmitters can be taken out, it's a lot easier to build some new Loran transmission gear and radio towers than it is to launch new GNSS satellites (which are also vulnerable to attack).
I think in reality, it’s easier to put little star trackers on your boats and planes.
And it’s way easier to launch another cruise missile at the rebuilt LORAN site.
3kg! That's a very high price to pay for reliability under GNSS denied environments in the day time. Surely you can get ~altitude from a barometer, ~position from inertial + IMU + prior position, and if you want more accuracy maybe DEM models + topography is better.
Therefore I'd suggest this is not going to be very useful on anything with a view toward efficiency (drones). This is more likely a 'quick fix' plugin for existing fat-plane avionics for 'identified risk' reasons. OK move commercially, but technically meh. The go to market plan would presumably be FUD around GNSS denied situations. Reality: unlikely to be problematic in most cases (known flight path, high altitude, low GNSS-denied environment dwell time).
If you tend to operate near / around the yellow and red zones shown here:
* https://gpsjam.org
Like an airport in Estonia:
* https://www.cbc.ca/news/world/gps-interference-airlines-1.72...
That's the annual salary of of 2-3 airline pilots, which you pay year after year in OpEx. This is a one-time CapEx that will work for years and may allow you to fly (i.e., generate revenue) when otherwise you would be grounded:
* https://www.cbc.ca/news/world/gps-interference-airlines-1.72...
And, oh, yes, all ships before last century, too, of course. So it is more spiral than circle :)
https://www.beadelltours.com.au/lb_survey.html
* "the largest land-based test range in the western world"
https://en.wikipedia.org/wiki/RAAF_Woomera_Range_Complex
https://timeandnavigation.si.edu/multimedia-asset/nortronics...