Collision Avoidance: Coordination, Precedence, Dispute Resolution

[Tywin]

It was just a PR hit, bad for everyone. The way Wyler and Forbes jumped in was highly indicative

Wyler appears to be in the process of getting his head handed to him for that... PR stunts often backfire.

Well looks like he don't give up...

https://twitter.com/greg_wyler/status/1168988584818425857

Jesus can you imagine being in business with him?  Sleazeball.

I'm start to understand why Elon gone alone with the Starlink...

[envy887]

Economic incentives work a lot of the time.  In some cases, though, they lead to unintended consequences, so they have to be very carefully set up.

Who decides the costs for each organization?  It seems like this system would heavily incentivize each organization controlling a satellite to give the highest possible cost estimate for its avoidance, and that can lead to intentional or unintentional distortion and sub-optimal behavior.

Good point.
We could imagine that launch contracts and satellite procurement contract all have to be disclosed to a single third party (of course with the right confidentiality requirements) that would be tasked with calculating collision avoidance fees, but there are cases where this wouldn't even be feasible.
Starlink is a great example of that. Since everything is done in an integrated way, there is no launch contract or satellite building contract and this would at best be declarative.
Military or intelligence satellites would be another corner case (but they are a corner case for satellite avoidance anyway. With these you just have to assume that they will perform all necessary maneuvers themselves and not tell nor charge anyone since they don't want details about what they are doing to be disclosed).

You could also just say that you have to declare to a third party what you will charge per maneuver per kg of fuel at launch and have that be public data and that's it. Then if you don't want to maneuver or if you want to play a game of charging all the maneuver cost to other parties, you'll declare a high price and probably end up being the one that pays most of the time.
The downside is that small satellites that have very low maneuvering costs may try to make a benefit of this  because they are so much bellow bigger ones.

In the end, I think most of the cases are handled by the rule I added in my edit : If you are intentionally putting yourselves on a collision course with the intent of "collision trolling" other parties (this is why this second rule is essential), you'll end up supporting all the costs of subsequent maneuvers. So in most of the cases, either it's an accidental conjunction and it's very rare, so someone charging a bit more than they should isn't really a concern or if there is an orbit that ends up being shared, there will be an agreement before anything happens that would cover who pays what and who does what.

Another point in favour of a system that makes people share the costs comes from something ESA mentioned, which is that most avoidance maneuvers involve dead satellites. Which means that would put an economic incentive on deorbiting or avoid populated areas to park dying satellites. It might help reducing LEO clutter if done properly.

P.S. : I also like the "pay me what you'd charge" idea. Although again, you have to be careful with the risk of a small provider "collision trolling" big sats.

The cost to maneuver in LEO, particularly low LEO, is basically irrelevant. In the case of Aeolus, it uses twice as much propellant every day to make up for atmospheric drag as it did to avoid the Starlink.

Also refer to Matt Desch:

It's not a big deal. Only typically need to move a satellite a few meters a day in advance of conjunction. Doesn't shorten life appreciably of satellite. We've been at it since 2009 when good catalog data was made public (thanks, unfortunately to our SV33 collision from no data).

https://twitter.com/IridiumBoss/status/1168618129892761602

The orbital changes we are talking about here are extremely tiny. Aeolus only used 0.1 m/s of delta-v to effect a 350 m change in height on the other side of the orbit.

[Lar]

It was just a PR hit, bad for everyone. The way Wyler and Forbes jumped in was highly indicative

Wyler appears to be in the process of getting his head handed to him for that... PR stunts often backfire.

Well looks like he don't give up...
(tweet deleted)

That's the original tweet which he is getting lambasted for. Wyler tweets are probably a bit off topic except in passing.

[Asteroza]

There is no current analogue to ADS-B.

I think this bears more attention in the context of autonomous avoidance as allegedly practiced by Starlink and expected in future sats. We currently have AIS for ships and ADS-B for aircraft, which allow vehicles on collision courses sufficient notice of another vehicle's current vector and to a lesser extent intended path. It would nice to be able to have a small beacon for sats that does the equivalent, as that would form the basis of basic true autonomous avoidance on a vehicle to vehicle level (with terrestrial analogues of V2V signaling between cars being standardized so vehicles, autonomous or not, can share sensor data with each other in close proximity, such as indicating the environment around a blind corner).

A problem here is the ability to react to a self determined collision based on detected beacons and the onboard TLE model. Ships are slow, airplanes are faster, but can power their beacon farther so the reaction timescale still fits. Is it reasonable for sats to be able to detect the beacon of another and maneuver enough to not hit, assuming we really minimize their keep-out spheres? Electric thrusters are not usually high thrust so reaction time here matters. But then again the accuracy of the beacon indicated vector matters too (especially if it's sending out bad data due to GNC issues), which then pushes things on either active sensors locally, or increasing reliance on TLE data from exterior sources.

There's also the simpler problem of the type of movement needing to be defined by an industry accepted algorithm, similar to how TCAS stuff works for airplanes. Does the sat on the ascending node move up/north, and the one on the descending node move down/south? If both are ascending, northenmost moves up while the southernmost moves down? Plus the algorithm should factor in other data provided by beacon, such as indicated mass and available thrust/deltaV as generally lighter vehicles are easier to move.

[Alastor]

The cost to maneuver in LEO, particularly low LEO, is basically irrelevant. In the case of Aeolus, it uses twice as much propellant every day to make up for atmospheric drag as it did to avoid the Starlink.

The orbital changes we are talking about here are extremely tiny. Aeolus only used 0.1 m/s of delta-v to effect a 350 m change in height on the other side of the orbit.

The cost of maneuvers may be irrelevant now, but we also have to consider also the future.
One thing is for sure, which is that the number of objects in space is about to raise quite sharply with both the development of the small sats market and the launch of several mega constellations.
Not only this, a lot of actors are entering the market.

With few actors, it's not very hard to make sure that all actors will act properly and with a relatively low number of sats, the number of encounters is relatively low, so it's not that relevent. as you point out, especially compared to drag in very low orbits.

However, once you multiply the sats in orbit by a number of 10 or so, and you add many actors, you significantly increase the risk of having to maneuver very often if you are not careful, disrupting service and having these insignificant costs add up to something that suddenly becomes significant. Especially if some of these new actors are not as carefull as they should about other actors, which you could easily see happening with some actors having to take care of several thousands of sats, where others have a few to a few dozen satellites.
If you push this even further, you could imagine someone trying to mess with their competitors by intentionally putting some sats in a disrupting orbit.

I think the worry expressed in the past few days comes from people looking at the curves of the number ofobjects in space and saying "yeah, if we multiply the number of daily encounters by the increase in sats we can expect, that would definitely become a problem".
And I'd consider this worry to be justified.

A regulation could be overkill, but also can be expected to prevent big problems for some users of space, so it's probably worth it.

[envy887]

The cost to maneuver in LEO, particularly low LEO, is basically irrelevant. In the case of Aeolus, it uses twice as much propellant every day to make up for atmospheric drag as it did to avoid the Starlink.

The orbital changes we are talking about here are extremely tiny. Aeolus only used 0.1 m/s of delta-v to effect a 350 m change in height on the other side of the orbit.

The cost of maneuvers may be irrelevant now, but we also have to consider also the future.
One thing is for sure, which is that the number of objects in space is about to raise quite sharply with both the development of the small sats market and the launch of several mega constellations.
Not only this, a lot of actors are entering the market.

With few actors, it's not very hard to make sure that all actors will act properly and with a relatively low number of sats, the number of encounters is relatively low, so it's not that relevent. as you point out, especially compared to drag in very low orbits.

However, once you multiply the sats in orbit by a number of 10 or so, and you add many actors, you significantly increase the risk of having to maneuver very often if you are not careful, disrupting service and having these insignificant costs add up to something that suddenly becomes significant. Especially if some of these new actors are not as carefull as they should about other actors, which you could easily see happening with some actors having to take care of several thousands of sats, where others have a few to a few dozen satellites.
If you push this even further, you could imagine someone trying to mess with their competitors by intentionally putting some sats in a disrupting orbit.

I think the worry expressed in the past few days comes from people looking at the curves of the number ofobjects in space and saying "yeah, if we multiply the number of daily encounters by the increase in sats we can expect, that would definitely become a problem".
And I'd consider this worry to be justified.

A regulation could be overkill, but also can be expected to prevent big problems for some users of space, so it's probably worth it.

It's not the number of active satellites in orbit that is relevant. It's the total number of objects, which is already near 20,000. As long as we enforce the guidelines on orbital lifetime and active disposal at EOL, the number of objects won't increase that much, perhaps 50 or 60% based on the current megaconstellation plans. This will not have a major impact on the number of conjunctions (50% more than rare is still pretty rare) nor on propellant budgets for conjunction avoidance maneuvers (50% more than trivial is still pretty trivial).

Doubling the number of active objects actually doesn't increase the conjunction avoidance maneuver propellant budget per satellite at all, because only half the satellites have to move in an active-active conjunction.

The key things to focus on are communication between operators, and enforcing end-of-life deorbiting.

[gongora]

Is it reasonable for sats to be able to detect the beacon of another and maneuver enough to not hit, assuming we really minimize their keep-out spheres?

I don't think that would be reasonable.  Sats carrying positioning payloads and communicating good positions to the ground for conjunction analysis is far easier.

[OTV Booster]

The following if a clumsy cut n paste from the StarLink forum. Fits better here.

Quote:
In rethinking several pucker situations I’ve had in choppers the immediate reflex is for the low craft to go lower and the high guy to go higher. The aircraft to the right breaks right and the left breaks left although differences in turn performance can make this sub optimal but better than nothing. One will dive and break one way and one will climb and break the other way. The only situation not covered is dead on head to head which if you have very precise tracking data will ALMOST never happen. For this a random generator throwing both craft in a random direction reduces the odd to some non zero but VERY VERY low chance of collision. Repeat random as needed.

Seems like a starting point for traffic rules.

Phil


It's not the same.   The time and distance scales are completely different. The warning is the result of a mathematical calculation about an object that you can't see, thousands of km away with many minutes or days of warning.  You know what it is and where it will be (or you don't know there will be a collision at all.)  Reflective action is not required.  You can take minutes (possibly thousands of minutes) to figure out the best course and try to communicate.  If you can communicate or the object has characteristics or intentions listed in whatever data base is being used to foresee the collision these can be taken into account. If you get to the pucker point with an unknown object looming in a view port you've already lost. Computers don't have reflexes in anyway comparable to humans.  Trying to get them to act like humans is very difficult, it's usually much easier to use a different solution that plays to their strengths, such as using a detailed checklist thousands of items long requiring millions of calculations.  The last few items on the list might resemble the actions you suggest, but these should not be the meat of the algorithm.


End quote


Cut me some slack guy. I may not be a physicist but I have heard of some of the work done by that Sir Issac feller. 
   
What you describe quite well is the situation preceding a decision to maneuver and the lead time needed to imitate action. I was proposing a possible algorithm for that action.

As the situation develops over time I would expect he ‘solutions’ to resolve down from two fuzzy areas (with time being one of the dimensions) with some degree of overlap into two sharper areas with more or less overlap which in turn represents the odds of collision. However far ahead of time it must be initiated, what I proposed, could form a basis for an automated avoidance algorithm. Has less ambiguity than ascending and descending nodes.  Because it does not depend on communications  it also works if one object is inert.

Yeah, if you can see em, you got more than just pucker.

Phil

[Tywin]

Old article from SN that is very relevant now...


https://spacenews.com/muddling-through-space-traffic-management/

“Space traffic management” is the cyber security of the space world. It is a hugely important public policy issue that underpins the successful future expansion of space activities, yet there is no agreement on its definition.

First, there needs to be a decision on whether the U.S. military will continue to provide all of the space situational awareness (SSA) data and services for the world, a task they took on after the 2009 collision between the Iridium 33 and Cosmos 2251 satellites. But the growing demands and complexity of the safety of spaceflight mission are increasingly competing for resources with the national security mission of the U.S. military, as multiple senior leaders have pointed out. Instead, it’s time to separate the civil SSA functions for safety of spaceflight from the core national security SSA functions, and assign the former to a civil federal agency.

However, the reality is that ICAO was created to resolve differences between national regulatory frameworks and relies on national administrations to implement and enforce its standards. Thus, it is extremely unlikely an “ICAO for space” would be successful before the existence of any national regulatory frameworks for space traffic management.

[Tywin]

Interesting news about this future problem...

https://www.esa.int/Safety_Security/Space_Debris/Automating_collision_avoidance

https://spacenews.com/u-s-space-command-eager-to-hand-over-space-traffic-duties-to-commerce-department/

Commerce requested $10 million for the Office of Space Commerce to create a new organization for space traffic management starting in fiscal year 2020. The government is currently operating with temporary stopgap funding and it’s not clear when Congress will pass appropriations bills for federal agencies.

https://www.technologyreview.com/s/614693/how-the-worlds-biggest-gun-helped-solve-a-long-standing-space-mystery/

[su27k]

How to better manage space traffic: Aeolus/Starlink encounter shows emails and late-night phone calls no longer cut it

Constellations, though, may not be the biggest threat in LEO. Darren McKnight of Centauri said the bigger threat comes from hundreds of upper stages, primarily Russian, left in three “clusters” in low Earth orbit at altitudes between 775 and 975 kilometers. The stages are large, with masses of up to 8,000 kilograms, and can’t maneuver.

The threat is not theoretical. McKnight said that, in May, two of those stages passed within 87 meters of each other at a relative velocity of 14 kilometers a second, missing a collision by just seconds. Had they collided, he said, “it would have doubled the catalog population in one event.”

This is why ESA's Aeolus announcement is so irresponsible, they focused the public on entirely the wrong threat, we have multi-ton un-controlled stages playing Russian roulette with each other at high orbit, yet they want the public to focus on hundred kilogram actively controlled satellites in low orbit. One has to wonder why, I guess it's much easier to intimidate a private company than a major nuclear power.

[john smith 19]

Interesting news about this future problem...

https://www.esa.int/Safety_Security/Space_Debris/Automating_collision_avoidance

https://spacenews.com/u-s-space-command-eager-to-hand-over-space-traffic-duties-to-commerce-department/

Commerce requested $10 million for the Office of Space Commerce to create a new organization for space traffic management starting in fiscal year 2020. The government is currently operating with temporary stopgap funding and it’s not clear when Congress will pass appropriations bills for federal agencies.

https://www.technologyreview.com/s/614693/how-the-worlds-biggest-gun-helped-solve-a-long-standing-space-mystery/

That MIT article is eye opening, and not in a good way. 

I'd long suspected the major threat was going to be the stuff you can't see IE sub 10cm, although I didn't realize there is some visibility of stuff in LEO to 3mm.

But there's so damm much of it.

All that carbon fibre and multi layer insulation has done no favors in terms of MMOD.