Sodern launches Astradia, a star tracker for GNSS-denied navigation

https://sodern.com/fr/astradia-le-viseur-detoiles-diurne-pour-un-systeme-de-navigation-sur-et-robuste/

by sorenjan • 1 year ago

27 47 1 year ago

blacklion2• 1 year ago

So, technology closed the circle. Nice.

ck45• 1 year ago

Those systems are not new, the SR-71 used the same principle, see e.g. https://www.iancollmceachern.com/single-post/unveiling-the-s...

blacklion2• 1 year ago

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 :)

nickspacek• 1 year ago

Neat, this latest implementation seems a little more compact than the SR-71's, but it's hard to tell.

https://timeandnavigation.si.edu/multimedia-asset/nortronics...

AStonesThrow• 1 year ago

Wow! A daytime star tracker! What wavelengths does it use?

stogot• 1 year ago

The e day part surprised me. I was familiar with other systems that use stars in darkness (very high altitude)

rich_sasha• 1 year 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.

sorenjan3• 1 year 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.

throw0101b3• 1 year 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:

* 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.

WJW• 1 year ago

I was wondering about this too and did some googling. https://iopscience.iop.org/article/10.1088/1742-6596/1510/1/... indicates that it's probably above 900nm wavelength, as beyond that you get only minor interference from refracted solar radiation.

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.

voxadam2• 1 year ago

English: https://sodern.com/en/astradia-a-daytime-star-tracker-for-a-...

sorenjan3• 1 year ago

That's strange, I submitted the english version but I see it has been changed to the french.

voxadam2• 1 year 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.

perihelions5• 1 year ago

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.

sneak2• 1 year 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.

wongarsu• 1 year 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

perihelions5• 1 year 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.

teamonkey2• 1 year 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.

The SR71 used a similar system.

https://theaviationgeekclub.com/the-sr-71-blackbird-astro-na...

perihelions5• 1 year ago

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.

throw0101d6• 1 year ago

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:

* https://www.npr.org/2016/02/22/467210492/u-s-navy-brings-bac...

contingencies2• 1 year 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).

shellfishgene• 1 year ago

Also a bit pricey at 250k Eur...

throw0101d6• 1 year ago

Also a bit pricey at 250k Eur...

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...

constantcrying• 1 year ago

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.

contingencies2• 1 year 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.

throw0101d6• 1 year ago

Use case = ?

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...

throw0101b3• 1 year ago

Someone is working on restoring an astro tracker from a B-52 bomber:

* 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/

aredox• 1 year 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.

https://ieeexplore.ieee.org/document/9179736

https://openstartracker.org/

throw0101b3• 1 year ago

I'm curious to know why resurrecting (e)Loran seems to be a hard sell:

* 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...

sorenjan3• 1 year ago

There's BPS (Broadcast Positioning System) too.

https://www.nab.org/bps/

relaxing3• 1 year ago

Because despite what they say the signals can still be jammed. Also your transmitters are massive targets for kinetic fires.

throw0101d6• 1 year ago

Of course any signal can be jammed if you throw enough watts out, but Loran is five millions times stronger:

* 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).

voxleone• 1 year 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.

throw0101d6• 1 year ago

From four months ago, "Celestial Navigation for Drones":

* https://news.ycombinator.com/item?id=42767797

hollerith• 1 year ago

This could help North Korea and Iran develop ICBMs.

Havoc2• 1 year ago

US had this tech 50+ years ago (sr71) so would think it’s within reach of most state actors now even without help

Havoc2• 1 year ago

Let’s hope they stick it on the airliners flying in the baltics

Sodern launches Astradia, a star tracker for GNSS-denied navigation

Related Stories