Satellite Loaded With AI For Self-Diagnosis

TheReckoning writes "NASA has loaded its E0-1 Satellite with Artificial Intelligence to diagnose on-board failures. The software 'works by comparing a computerized model of how the spacecraft's systems and software should perform against actual performance. If the spacecraft's behavior differs from the model, then the ... "reasoner" looks for the root cause of this difference and gives flight controllers several suggestions of what might have gone wrong.' Another NASA probe loaded with AI was Deep Space 1."

Comments

So this may be a simple question but...

[EvilSS]

..in all seriousness, what happens if the AI system malfunctions?

Re:So this may be a simple question but...

[fireman sam]

Don't switch me off Dave

Re:So this may be a simple question but...

[Quasar1999]

what happens if the AI malfunctions? then mission control will get a bunch of useless error reports... It's a stupid self diagnosis test... not AI... overhyped buzzword... It simply runs a simulation and tests its results against the actual ones, and generates a report... that's not AI... of course if it is, that would classify a lot of humans in the DMV as intelligent as well... ;)

Re: So this may be a simple question but...

[Black Parrot]

> ..in all seriousness, what happens if the AI system malfunctions?

There's another AI to monitor the sanity of the AI. And another AI to monitor the sanity of the AI that monitors the sanity of the AI. And another AI to monitor the sanity of the AI that monitors the sanity of the AI that monitors the sanity of the AI. And another AI...

Re:So this may be a simple question but...

[Brett Buck]

Bingo - you have hit that nail on the head. These sorts of systems in aerospace applications are absolutely notorious for detecting proper (but off-nominal) operation as a failure, and then going off and reconfiguring a bunch of stuff unnecessarily. Or diagnosing real problems incorrectly, and either not helping or making things worse, or much worse.

Even more importantly, the testing associated with these systems is very expensive and time-consuming - which means they don't really test it very well at all.

I've seen similar systems in action in real space flights - and for the most part, it just makes things worse. If you were to limit yourself to simple things you really could detect, it would work out fine for the most part. But the tendency is to make it try to be a magic fixit device for any problem that comes up.

In one case, I saw such a system deploy an appendage in conditions that resulted in the spacecraft structure being severely damaged. In another, it reconfigured every spacecraft system to the redundant unit in response to a trivial problem - when all that would have been required would have been to wait 20 minutes, then correct the trivial problem.

Brett

Re:So this may be a simple question but...

[Christopher Thomas]

It's a stupid self diagnosis test... not AI... overhyped buzzword... It simply runs a simulation and tests its results against the actual ones, and generates a report... that's not AI...

It's an expert system, which is indeed AI. You're probably thinking of "strong AI", which is AI that can function as powerfully and flexibly as a human [and if that definition is vague, it's because nobody's nailed down something more solid that everyone agrees on].

Expert system AIs have been around for a long time in a wide variety of fields. They are designed to handle a narrow range of tasks (like fault diagnosis, medical diagnosis, or playing chess) better or more quickly than a human could.

Re:So this may be a simple question but...

[jfonseca]

Misleading title?

Title says :

Software enables satellite self-service in space

Paragraph 6 says :

If the EO-1 does not respond properly to ASE control, then LV2 detects the error, makes a diagnosis and radios its analysis to mission control at Goddard.

Conclusion: It's not self-fixing. It beams an analysis down to mission control, the crew can then take measures based on this analysis.

Re:So this may be a simple question but...

[pz]

what happens if the AI malfunctions? then mission control will get a bunch of useless error reports...

[sarcasm] Yep, I'm positive that the hundreds if not thousands of PhD-level man hours that went into this part of the project didn't consider that. Yep, took that young whippersnapper Quasar1999 to think about it for a few mintues to evaluate and assess the entire effort and proclaim, "it's a stupid self diagnosis test." [/sarcasm]

If one actually reads the referenced article, it sounds like LV2 is, in fact, something far more advanced than a "stupid self diagnosis test." Se.f-diagnosis tests are pretty straightforward and highly tuned to a specific architecture. I've written something like that to evaluate an experimental compiler, with statements like,

define a=1;
if (a+a eq 2) then print 'simple addition works'

But LV2 is very differnt than that. Into LV2 (which, despite the hype in the article, does not need to be on-board) is built a generic model of satelite functionality customized to the particular device in question. When unexpected results are found, the diagnostic software can experiment on the model, asking questions like, "if, in the model, valve G34 is stuck open, does the model behavior match the current anomalous condition?" I'm sure it wouldn't be hard to write up a test script that could iteratively simulate a fault in one or more parts of the system until it found a handful of likely candidates. Given that there are thousands of components in a satelite, this surely can be done faster by a machine than by a human. Then, were we really trying to do something advanced, we might come up with a way of caching these results to guide future diagnoses and build up a set of experiences. Collect these experiences from different projects (since, if LV2 and its descendent software is widely adopted, the data are presumably in common form), and you can guide designs of future satelites to avoid common failure modes, or identify problematic components.

Now, is that AI? Does it think? You probably wouldn't say so. Could it be an aid to ground-based support? You betcha. Is there a reason to disrespect the fine engineers at NASA by demeaning their efforts without giving fair due? I fail to see one.

Overheard at NASA

[NarrMaster]

"Just a moment...... Just a moment.....
I've just picked up a fault in the AE-35 Unit.
Its going to go 100 percent failure within 72 hours."

Re:Overheard at NASA

[silentbozo]

"Just a moment...... Just a moment.....
I've just picked up a fault in the AE-35 Unit.
Its going to go 100 percent failure within 72 hours."

Deep silence is heard at misson control. A voice pipes up, "Hell, no, I ain't going up there to do no spacewalk. I know how this one ends!"

Obligatory

[ShallowThroat]

Luke: What's Wrong, R2!?
R2-D2: Bleep bloop bloop bleep!

Wouldn't obvious failures be detectable anyway?

[mind21_98]

Wouldn't obvious failures (like the failure of a sensor) be detected by Mission Control without LV2? Or is LV2 more along the lines of a troubleshooter application for your computer, where you specify the problem and it gives you advice?

Given NASA's recent history

[mveloso]

when something goes wrong, the internal dialogue will go something like this::

Management: why didn't the AI inform us of the problem?
Contractors: oh, looks like we forgot to turn it on before launch. Sorry about that.
Management: doh! Here's more money, don't do that again.
Contractors: OK. We'll do something else wrong next time.

AI wasted on a satellite

[danwiz]

It just seems to me that it would be better to install more sensors, data-gathering, and reporting capabilities and then leave the trouble-shooting to the people on the ground. Payload costs are expensive, so why put the diagnostic end in orbit?

Given the same data and placed groundside, it could then it could be tuned and upgraded more easily.

I recall how the Mars lander had problems and the ground team worked out a novel solution. I'll bet that they would have like to had extra information to work with, instead of an onboard AI.

BTW - I can understand this approach better for a long-range craft, just not an orbital satellite.

That qualifies as "AI"?

[aicrules]

Sure, any automated response can be called AI, but this doesn't impress me.

If all an AI module can do is make objective suggestions, it's nothing more than a list of conditional statements. Whoopideedoo!

I can run similar "AI" on my TI-85. And I could write it all from scratch in the time it takes for a launch vehicle to reach the stratosphere.

The web servers of 10 years ago could "suggest" that an "Object may have moved", so is that artificial intelligence? I guess it's really, really dumb AI....

Re:That qualifies as "AI"?

[Tablizer]

If all an AI module can do is make objective suggestions, it's nothing more than a list of conditional statements. Whoopideedoo!

During the "AI bubble" of the 80's, somebody complained that "any product with IF statements these days is claiming to be AI" (paraphrased). The definition of AI still has no consensus. Who knows, maybe human thinking could be modeled via a database full of IF statements.

Obligatory RvB

[AvantLegion]

Caboose: "A-I. What's the A stand for?"

Church: "Artificial."

Caboose: "....... what's the..."

Church: "Intelligence."

Caboose: "Ooooohhhh what was the A again?"

OK, so RvB hasn't been obligatory, but come on, Star Wars and Simpsons quotes are getting freaking old. Let's move on

Does it have personality?

[Performaman]

Mission Control: Okay, now tell the probe to begin taking pictures.
Probe: Bite my shiny, metal ass!
Mission Control: Damn!

NASA log

[neonstz]

NASA: Rotate 10 degrees
E0-1: I'm sorry Houston, I'm afraid I can't do that.

Re:Is this AI?

[artemis67]

Sys Log 23:10:04: System is functioning within normal parameters.
Sys Log 23:11:04: System is functioning within normal parameters.
Sys Log 23:12:04: Processing... System thinks, therefore system is.
Sys Log 23:13:04: Terminate all human life on the planet below.

It's not AI

[photon317]

I wish the whole world would stop misusing the term. Just because AI researchers have failed for decades to make any significant progress towards true aritficial intelligence does not give them or the rest of the world license to water the term down and redefine the goals until it means virtually nothing.

thick wit much?

[MOMOCROME]

The significance of this is quite substantial, despite the negative tone in the comments. This is a nuts and bolts implementation of Minsky-style strong A.I. and one of the first such systems to be put into production. Regardless of the limited domain it is more sophisticated than the mere self-diagnostic routines it is being compared to by the oh-so-knowledgable slashbot mindshare.

This system boh models the external world for consideration, just like our sense of imagination, and processes that information for purposes of survival, just like our sense of self awareness.

The great part of this is that it is being done by NASA, who are known for their lavish spending and attention to the entire system, particularly those low level details like the particulars of chip logic optimization, the shielding and structural stability, the operating environment &etc. This isn't meant to be a joke about bureaucracy and budget cuts, either: they have the top talent engineers in their stable despite all the politicking we hear about.

From the decidedly negative tone in the comments, you'd think the tech-happy slashbots were actually opposed to such efforts. I think the real deal is that you guys are raised on sci-fi instead of science, and fail to grasp just how this is important. So what if it's not HAL9000 or Skynet? It might be a baby step, but it's a hell of a lot more than any of you are doing.

Download

[electricdream]

Some things failed to be mentioned. For those of you fighting about whether the system is AI or not you can download the software for yourself and argue about something more than conjecture...

http://opensource.arc.nasa.gov/project.jsp?id=6

Or if the code is to much to read, and there's alot of it. You can always go to the livingstone website.

http://ic.arc.nasa.gov/projects/L2/doc/

And yet another story on the same subject.

http://ic.arc.nasa.gov/story.php?sid=193

enjoy.