Japanese 'Minerva' Robot Lost in Space

space_weasel writes "A little Japanese robot that was supposed to land on the surface of an asteroid has accidentally been flung into space by its mothership. New Scientist Space reports that the accident occurred as the data link with the spacecraft was being switched from an station in Japan to one in Australia. Mission controllers still plan to punch a hole in the asteroid and collect samples, which will be returned to Earth."

Comments

Whoops!

[Skye16]

I wonder how, exactly, the software being used had the capability to allow this to happen. Even if the problem were procedural, I would think that, on transfer of control, you would lock down all non-essential functions - like "flinging" payloads into space - until control has been successfully handed off.

Of course, this is all pointless conjecture on my part - it may have been a hardware malfunction, for all I know. It would be interesting to analyze things like these. Having only a few years real-world experience, I doubt my programming skills would be worth a damn, but I would be thrilled just to have the opportunity to read the code they use before hand. Generally I don't volunteer my time to OSS-like programs, but this is one situation where I could easily see myself helping. Or trying to help, more like it.

Then again, by releasing it beforehand open source, someone else may very well be able to analyze the code and "steal" control of the probe/satellite/whatever-is-using-the-software, possibly using it for nefarious gain, or possibly just being a bunch of dicks. So this probably wouldn't pan out. Still, a nerd can dream.

Re:Whoops!

[Otter]

Even if the problem were procedural, I would think that, on transfer of control, you would lock down all non-essential functions - like "flinging" payloads into space - until control has been successfully handed off.

As I understand the story (which could easily be wrong), they had to issue the release command blindly, because the need to make the adjustment came up precisely when Murphy's Law predicts. Having the flexibility to do so at least allowed them to make the gamble that they wound up losing.

That's why we need AI in space

[MOBE2001]

During this 40-minute antenna change, information about the spacecraft's vertical motion was unavailable to ground controllers.

For a country which prides itself as being at the forefront of robotics technology, this is rather surprising. The latency inherent in space communication over great distances is the primary reason for using intelligent robots in space. If the probe was sufficiently intelligent, it would perform its tasks without supervision from ground control. I hope they (including NASA and the ESA) put a lot more effort into automating their space probes in the future.

It was hardware and circumstance

[ianscot]

Of course, this is all pointless conjecture on my part - it may have been a hardware malfunction, for all I know.

You'd sure know more if you went to the (somewhat unclear) article, which would obviate the need for lots of your conjecture.

The main probe has been going on one of its three "stablizing wheels," the other two having failed. There's a sidebar link in the article to an earlier one about those failures. Mission controllers have been burning extra fuel keeping the thing at the right distance from the asteroid, facing the asteroid, and with its solar panels facing the sun; they already had that against them. Then the altimeter data they were getting was bad, they were closer than they thought, because some combination of the laser altimeter (previously untested) and the slope of the asteroid's surface confused the data.

They realized they were within 100 meters and had to send the detach command while the antenna switch was happening. The blackout prevented them from realizing a "keep above minimum altitude" engine thrust had just gone off.

This is much more of a reflection of this model of probe: it's cheaper, it's faster to develop, and there are going to be failures like the Beagle and this.

(Personally I do think there'd be a big gain if, before and after missions like this, the code got reviewed. I doubt very much that hackers in Idaho would have foreseen the failed stabilizers, the workaround, the potential for misjudging the altimeter data, and the combination of the blackout and the necessity for the release command. But in terms of intellectual freedom, it'd be a nice statement, and the Post Mortems would sure feature a lot of people asking Feynman-esque questions about icewater and O-rings.)