Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[RedKing]

("Orbitar" looks a bit proto-"Planety McPlanetface", doesn't it?  )

It's interesting, because apparently it's a contrived name in reference to NASA's orbital operations. I don't quite get why that ended up being "Orbitar" as opposed to "Orbiter" or "Orbita", but there you have it.

It's spacey. Pulsar, Magnetar, etc. The 'tar' is from "star" I think, so it's technically incorrect (it'd be a better name for the sub-star in a dom/sub binary system, an "orbiting star") but whatever.

Descan, please keep your dong out of my stars.

His is the dong that will pierce the heavens.


So, NASA is working on designing a troop of robots called Valkyries to serve as an advance party for colonizing Mars.

[Sirus]

("Orbitar" looks a bit proto-"Planety McPlanetface", doesn't it?  )

It's interesting, because apparently it's a contrived name in reference to NASA's orbital operations. I don't quite get why that ended up being "Orbitar" as opposed to "Orbiter" or "Orbita", but there you have it.

It's spacey. Pulsar, Magnetar, etc. The 'tar' is from "star" I think, so it's technically incorrect (it'd be a better name for the sub-star in a dom/sub binary system, an "orbiting star") but whatever.

Descan, please keep your dong out of my stars.

His is the dong that will pierce the heavens.


So, NASA is working on designing a troop of robots called Valkyries to serve as an advance party for colonizing Mars.

Using humanoid robots on a distant planet sounds like a terrible idea, considering how unstable they continue to be even in the best of conditions. If one of them faceplants and can't get back up, that's several million dollars wasted.

What's wrong with good old wheels?

[TempAcc]

This just in: Mars will be conquered for humanity by a team of 4 ugly power rangers.

In all seriousness, I'm with Sirus on this. We haven't exactly managed to make full bipedal workable robots on earth yet, why are they considering sending 4 of them to mars to operate on uneven, rocky, dusty terrain?

[Egan_BW]

I guess the reasoning is that rovers wouldn't be very good mechanics.

[MetalSlimeHunt]

That's part of it, yeah. The human form is kind of wobbly when not controlled well, but don't forget that our shape provides an extreme level of versatility. We evolved into this form as a consequence of needing to do...all the things that we do, and it's unlikely that a non-humanoid robot of similar capacity would be able to match a humanoid robot of similar capacity.

For building structures on Mars, that's the kind of versatility and ability we absolutely need. That goes double if what they build is ever inhabited by flesh-and-blood humans, I can just imagine the kind of unforeseen ergonomic failures two-foot tall balls of wheels and welding torches would construct. We'd also have to build a couple hundred more of those than these, thanks to their specialization.

All in all, I get the concern after years of watching clumsy soccer robots, but let's not lose sight of NASA's job being to make the big leaps forward like these. The goal here is to essentially make a robotic copy of the human form. Humans who faceplant can stand up again. So long as they do the job, it'll be fine.

Windy. Don't forget windy.

Mars isn't windy. Or at least not meaningfully windy, as the air pressure is so low even a massive gust can't exert force on the surroundings.

[RedKing]

This just in: Mars will be conquered for humanity by a team of 4 ugly power rangers.

In all seriousness, I'm with Sirus on this. We haven't exactly managed to make full bipedal workable robots on earth yet, why are they considering sending 4 of them to mars to operate on uneven, rocky, dusty terrain?

Not too worried about that.

[TempAcc]

Didn't google put up boston dynamics for sale not too long ago?

[Sirus]

@ MetalSlimeHunt: I was actually thinking more along the lines of Curiosity-sized robots. That thing was about the size of a small car, IIRC. Much bigger than most robotic rovers, and with specialized arms I'm sure they could put together pre-fabricated structures.

This just in: Mars will be conquered for humanity by a team of 4 ugly power rangers.

In all seriousness, I'm with Sirus on this. We haven't exactly managed to make full bipedal workable robots on earth yet, why are they considering sending 4 of them to mars to operate on uneven, rocky, dusty terrain?

Not too worried about that.

It's getting there, no question, but still a bit unstable for my tastes. At least if we're talking about sending it several light-minutes away with absolutely no chance of recovery.

[RedKing]

Didn't google put up boston dynamics for sale not too long ago?

Huh....yeah, apparently they did. Based on some internal emails, even Google execs are a bit uneasy at the questions raised by BD's robots potentially putting people out of work. And slaughtering us all when they become the blood-slicked, hydraulic arms of the Google overmind.


@Sirus: True, but look at what we sent to the moon with people inside. Sometimes ya just gotta trust your engineers and cross your fingers. Not saying it's there yet, but I like this approach of robotic advance teams.

[TempAcc]

I'm pretty sad about the whole BD thing, but I doubt google's execs actualy did so because they're omg scared of the robots. Its probably because they couldn't think of a way to make money out of it anytime soon.

Anyway, there's a few unanswered questions about bipedal robots on mars, too. Would their stabilization system work on martian gravity? I mean, their systems are calibrated to handle earthly gravity, would operating in lower gravity affect it? I mean, when the austronauts got to the moon, even they had some trouble moving around and keeping themselves balanced while moving. Granted, moon gravity is significantly lower then martian gravity, but there's still a difference.

[MetalSlimeHunt]

I'm sure you could do testing for one-third gravity conditions.

[RedKing]

I'm pretty sad about the whole BD thing, but I doubt google's execs actualy did so because they're omg scared of the robots. Its probably because they couldn't think of a way to make money out of it anytime soon.

Anyway, there's a few unanswered questions about bipedal robots on mars, too. Would their stabilization system work on martian gravity? I mean, their systems are calibrated to handle earthly gravity, would operating in lower gravity affect it? I mean, when the austronauts got to the moon, even they had some trouble moving around and keeping themselves balanced while moving. Granted, moon gravity is significantly lower then martian gravity, but there's still a difference.

Thanks to the magic of physics and math, you can calculate a lot of that into software, and adjust on-the-fly.

Plus, I'm guessing they can do simulated 0.38G testing in water chambers and Vomit Comets.

[TempAcc]

Thing is, as much as I'm a Bostrom fan, being SUPER SCARED of AI right now isn't that great of a reason to give up on any sort of investment like this. We aren't even close to functional general AI, much less one that functions at a human level of adaptability, with super intelligent AI being still a thing we can only dream of (for now). I do expect to see some sort of working general purpose AI within my lifetime, but certainly not in the next 10 years, unless someone makes a huge breakthrough of some sort.

AI that is capable of learning doesnt necessarily mean it is also capable of improving itself past the boundaries of its programming, this is why we currently have some really good specific task AIs (like the alphago one), but no real general purpose AI. Nobody has even come up with a prototype for a generalistic AI yet, AFAIK.

[MetalSlimeHunt]

Despite being in the transhumanist crowd, I don't and have never bought the threat of a maladapted AGI. The logistical bottlenecks alone would prevent apocalypse even if we ever solve the Chinese Room.

[Starver]

Thanks to the magic of physics and math, you can calculate a lot of that into software, and adjust on-the-fly.

My guess is that they'd be allowed/made to relearn some of the behaviour they already learnt for 1g conditions. Without having poked at BD's press releases at all, I've a feeling that a lot of the 'stagger recover' from both deliberate pushing and awkward terrain conditions was 'learnt', not painstakingly programmed in, except for maybe some initial guidance as to how to narrow down the behavioural search-space towards an anthropocentric expectation.  The 'press-up' recovery from having fallen flat on its face, if tried in lower-g, would potentially send it crashing over onto its back, until it learns to push slightly lighter/slower for Mars-gravity.

Plus, I'm guessing they can do simulated 0.38G testing in water chambers and Vomit Comets.

Proof of concept that they are versatile, but water-testing would involve an artificially over-thick 'atmosphere', and an actual learning algorithm might end up going for windmilling arms 'swimming' its way back upright, in a way that would not work in Mars's air... And if Earth-air thickness is at all exploited right now then that ununtentional reliance will have to be unlearnt too.

(And can they, or rather do they, use the VC in a flight plan that produces low-g that's still not actually (effectively) zero? As in, doubtless they can (just don't go with quite as extreme a curved flight), but how easy is it to maintain fractional-g without wobble, if they're mostly configured/pilot-trained to aim for zero?)


My thoughts are that a decent versatile robot to send as constructor would have a stable rover wheelbase (active suspension giving hexapodal-walking capability, built in for wheel-unfriendly situations) with torso/head/arms atop/in front for 'fine work' (autonomously or with remote telepresence by a human operator sat in a rig) and a crane-arm-cum-extendable-multiheaded-tool thing on the back (with carrying space between for building components/materials).  Add a retracting dozer-blade/stabilisation pad, and it could deal with many situations, even toppling over.. As long as it doesn't lose its ability to think and/or communicate.

(Counting up, there's at least 40 degrees of freedom, even ignoring end-effector/tool control, which is a lot of opportunities to go wrong, with a large number of non-interchangable components, but if it damages a wheel-motor it could limp back to base on five, four, even three remaining wheels or even actually limp home in slow-walk mode, or else buggyboy2 goes out to meet it with a spare part and effects a field-repair or helps tow it back.)