Humanoid Robots for Moon and Mars

[Cheapchips]

Somewhat reluctant to start this thread, as it's all really early, but then Moon and Mars missions are 3+ years away...

At Tesla's AI day they announced they are working on Tesla Bot, leveraging the work Tesla already does with computer vision, robotics and volume manufacturing.  A "we might as well do it, as we have all the pieces" move.

The shortage of unskilled/partially skilled Labour is a huge issue for Mars.  A ship of 'bots would be an extremely useful resource. They need so much less support mass than humans.  They could outnumber humans at some suitable ratio.

Doesn't sound like they've done that much work on it.  More like early physical design and a promise of exciting future work for existing and new AI engineers. 

If we do see a prototype demonstrating useful stuff in 2022 I'll be surprised. 

[DistantTemple]

I'm surprised SX or Tesla .... or Neurolink... probably not Dogleg park, but some Musk outfit isn't acquiring Boston Dynamics. Maybe they don't want to loose their identify, or to continue research in their own direction, however having access to some Musky AI, or the best chip design in the world should be a collaboration made in heaven!!!!
Also of course mass production expertise...
I keep expecting it!
Edit: As we have seen Spot is already applied to launch site monitoring, and Spot or Atlas etc or the technology applied in other ways is ideal for the Moon or Mars. We have seen NASA use Spot in lava tube mapping I think!

[Cheapchips]

I'm surprised SX or Tesla .... or Neurolink... probably not Dogleg park, but some Musk outfit isn't acquiring Boston Dynamics.

Hyundai purchase 80% of Boston Dynamics back in June.  You could argue that Tesla Bot is a reaction to that. 

[DistantTemple]

Just as Boston Dynamics is not relevant here, neither is Tesla Bot.

I suppose not, here on earth Tesla Bot, doesn't seem to be spaceflight related, however a version on the Moon or Mars, (see OP) does seem so.

[rakaydos]

Just as Boston Dynamics is not relevant here, neither is Tesla Bot.

I suppose not, here on earth Tesla Bot, doesn't seem to be spaceflight related, however a version on the Moon or Mars, (see OP) does seem so.

Sounds like room for a thread in Advanced Concepts. "Applications of Teslabot and Boston Dynamics in spaceflight."

[RoadWithoutEnd]

The shortage of unskilled/partially skilled Labour is a huge issue for Mars.

Why would there be a shortage?  Engineers and physicists are perfectly capable of turning a wrench, working a forklift.

They need so much less support mass than humans.  They could outnumber humans at some suitable ratio.

The best part is no part, so there would have to be a pretty extreme necessity argument for building a Mars colony with the most complex technologies available.  Especially if it would cause human colonists to avoid learning the key skills and survival strategies needed to make a truly independent civilization.

Places built with machines that couldn't be made locally for generations would be little more than tourist resorts, and by the time there were enough people to in-source production, there would be no incentive.  Everyone would be accustomed to lifestyles only made possible by being an extension of Earth's economy.

This would be Bad.

[geza]

If a task can be controllad by AI, like FSD in the (near?) future, the bot will not be needed. As an FSD computer can be embeded into the car without a bot, it will be possible to embed into an excavator, or into a manufacturing robot.

The bot is needed only, when it is convenient for humans to be served by a human-like object. Maybe, a humanoid waiter is preferred over a conveyor system providing the meal.

The usual hidden assumption here is that a "superintelligent" AI, is something like a human. This assumption is wrong in an essential way.

The AI of the foreseeable furure will be quite far from humans. What we call "intelligence" in AI context is just like e.g. the visual cortex of a human: it is programmed/teached to classify inputs and providing outputs accordingly. Humans are much more than this: humans are living beings, evolutionarily selected for self-replication. The visual cortext is just a tool for this. Other parts of the brains generate the goals as derivatives from the ultimate goal of reproduction vired into our DNA by evolution.

So, even if an AI chip surpass human brain in e.g. the number of "neurons", it will not be human-equivalents without internal goal generation. Of course, it will be possible to develop an artificial equivalent of the human brain, as a whole. However, it is much more than just a linear extension of the current AI along the deep learning paradigm.

A real brain is evolutaionary selected to serve the individual. So, an artificial equivalent of the human brain should be embedded into a real artificial organism, whatever it means. A bot developed around an FSD chip is not an artificial human.

[mrdje]

The shortage of unskilled/partially skilled Labour is a huge issue for Mars.

Why would there be a shortage?  Engineers and physicists are perfectly capable of turning a wrench, working a forklift.

They need so much less support mass than humans.  They could outnumber humans at some suitable ratio.

The best part is no part, so there would have to be a pretty extreme necessity argument for building a Mars colony with the most complex technologies available.  Especially if it would cause human colonists to avoid learning the key skills and survival strategies needed to make a truly independent civilization.

Places built with machines that couldn't be made locally for generations would be little more than tourist resorts, and by the time there were enough people to in-source production, there would be no incentive.  Everyone would be accustomed to lifestyles only made possible by being an extension of Earth's economy.

This would be Bad.

Peoples should stop throwing "best part is no part" everytime everywhere.


There are a lot of extremely difficults things to handle for people to be able to colonize mars.

Having robots coming before humans may be really handy to start building everything needed for humans - but not for robots-.

Living on Mars may be impossible without infrastructure to produce oxygen or water before even starting sending peoples on Mars... This peoples are going to have A LOT of things to troubleshoot and learn even if they're heavily helped by robots.

Colonizing Mars or the Moon may be the most difficults things ever attemped by humans... Even if they use every advanced technologies availables, I'm not even sure they're going to be able to achieve that.

[edzieba]

Humanoid robots are the right choice only in environments built for humans where it is hard to fit a specialised chassis.
Existing Earth buildings are built for humans. Mars is not. There is no good reason to use humanoid robots on Mars when you could use robots more suited to that environment.

[daedalus1]

Humanoid robots are the right choice only in environments built for humans where it is hard to fit a specialised chassis.
Existing Earth buildings are built for humans. Mars is not. There is no good reason to use humanoid robots on Mars when you could use robots more suited to that environment.

But the whole point is it will be built for humans.

[tssp_art]

Humanoid robots are the right choice only in environments built for humans where it is hard to fit a specialised chassis.
Existing Earth buildings are built for humans. Mars is not. There is no good reason to use humanoid robots on Mars when you could use robots more suited to that environment.

Well, I  mostly agree. But I can envision a humanoid robot that is more suited to the martian or lunar environment than humans are. Specifically with the ability to operate in vacuum in extreme thermal conditions for extended periods without fatigue or hunger. Given that here's just some thoughts on why that might be useful.

- A humanoid robot can perform a routine EVA in place of a human with less risk and perhaps much less fuss - it doesn't have to don and extensively test (and maintain and refurbish) an EVA suit).

- A humanoid robot can move unencumbered inside facilities designed for humans even if pressurization and/or thermal control is lost.

- A humanoid robot can use tools and operate equipment designed for humans.

And yes, one could design a robot that is more specialized and effective for specific missions needed for exploration and colonization. I fully expect a very smart rover for example as well as some very smart robotic construction and mining equipment.

But I also believe that there may well be a role for general purpose humanoid robots to amplify the capability of the astronauts/explorers/colonists.

[mrdje]

Humanoid robots are the right choice only in environments built for humans where it is hard to fit a specialised chassis.
Existing Earth buildings are built for humans. Mars is not. There is no good reason to use humanoid robots on Mars when you could use robots more suited to that environment.

- A humanoid robot can move unencumbered inside facilities designed for humans even if pressurization and/or thermal control is lost.

- A humanoid robot can use tools and operate equipment designed for humans.

THIS PART.

If you send humanoid robots on Mars they can start building the habitat for future humans AND navigate inside, even without the shielding / oxygen parts done. It would be way more difficult for wheeled rover.

Also they can use "regular" tools that can be also used by humans when they start arriving on Mars, you don't need specialized parts that are useless for humans, you avoid sending multiple times the "same" tool.

[su27k]

The shortage of unskilled/partially skilled Labour is a huge issue for Mars.

Why would there be a shortage?  Engineers and physicists are perfectly capable of turning a wrench, working a forklift.

They are capable, but if they're doing labor, then they're not doing engineering or science, which is the problem.

Labor shortage will be a big problem for Mars, I think Elon even mentioned it sometimes ("plenty of jobs on Mars", "needs to motivate enough people to go to Mars", etc), so I'm glad he's starting to address this.

Places built with machines that couldn't be made locally for generations would be little more than tourist resorts, and by the time there were enough people to in-source production, there would be no incentive.  Everyone would be accustomed to lifestyles only made possible by being an extension of Earth's economy.

That's going to be the case for Mars colony whether you use Tesla Bot or not.

The incentive for local production is cheaper price (transportation cost would still dominate even if you have Starship) and faster iteration (doesn't need to wait for 2 years to receive your newly designed chip).

[edzieba]

If you send humanoid robots on Mars they can start building the habitat for future humans AND navigate inside, even without the shielding / oxygen parts done. It would be way more difficult for wheeled rover.

Or you can design your environment for dual-use from the start, rather than hobbling your robots with anthropomorphism.

[Aussie_Space_Nut]

I have many years of experience with Robots. (CNC Machines)

They always have a human nearby to instruct/save them. (Needed Often)

Even all of Teslas robot cars have a human nearby to instruct/save them.

However

Elon said their Humanoid Robot will be able to go to the shop all by itself. If this happens it will be a game changer.
1) Will you have to take your robot to your local shop once, then it will be ok on its own?
2) Or perhaps another robot has already been to your local shop and all your robot has to do is download the details and then off it goes.
3) Or will your local shop have on it's wifi a download on demand map.
4) Or will you just tell your robot to go to the shop at such and such address and it is able to work it all out on the fly?

If case 4 is correct then the applications for Mars is huge. For Example. Go setup a solar array at a specific location. (The bot is able to go get the parts from storage, operate any machinery req'd, navigate to position and then set up the array.)

It can do all of this because the "Full Self Driving" (FSD) suite is able to identify the storage unit, it can identify the parts, it can identify & operate any machinery, etc.

The only catch and its a doozy! You would have to teach the robot how to assemble the solar array. Wiring it into the grid would also be a challenge.

I suspect that these humanoid robots will be a great aid on the surface of Mars when humans are present to instruct/save them.

I'm not so sure that they will be much help on Mars before humans arrive. As any hiccup in their operation will have to be dealt with remotely and that takes time.

But. Will they be able to simulate Mars operations using both the DOJO Supercomputer and perhaps a big sandbox somewhere such that they can greatly reduce the need for remote intervention prior to leaving Earth?

I think there is definitely potential here because FSD in Tesla Cars is working to some extent right now.

[electricdawn]

This may be a stupid question, but where are the batteries on that thing? If they're onboard, I'm not sure it'll last long enough to do anything useful, before it has to go back and charge itself up again. How long does one of the Boston Dynamics robo dogs last?

Thanks for letting me know.

[pjm1]

As any hiccup in their operation will have to be dealt with remotely and that takes time.

But. Will they be able to simulate Mars operations using both the DOJO Supercomputer and perhaps a big sandbox somewhere such that they can greatly reduce the need for remote intervention prior to leaving Earth?

Remote intervention is not a deal breaker IMHO when compared to the alternative: risking sending humans to complete the same tasks in the first place.  This point applies whether we're talking about humanoid robots or task-specific robots.

If there's an error in assembly/installation/whatever, as long as the robot can identify there's an error then waiting 20 mins for the error to be alerted to Earth and then whatever time for the problem to be solved and instructions sent back is hardly a big deal.  Absolute worst case, you have to wait until the next synod and send new robots with improved abilities and whatever solution to the infrastructure you need.

Certainly less of a big deal than sending humans and finding there's a potentially intractable problem with the infrastructure/setup and having to wait 26 months for a solution to be sent.

[Cheapchips]

This may be a stupid question, but where are the batteries on that thing? If they're onboard, I'm not sure it'll last long enough to do anything useful, before it has to go back and charge itself up again. How long does one of the Boston Dynamics robo dogs last?

In the chest, behind the FSD board by the looks of it.

Spot runs for ~90 minutes before it finds it's charger and tops up.  It's a 605 Wh battery.

Can't really say what that means for a produce still in development.  Tesla are obviously good at motors, batteries and power management.

[xvel]

And the spot is really unnecessarily heavy, very far from optimal.

[tssp_art]

So I'm thinking pretty simple tasks that are just hard to do in an extreme environment - an environment that takes planning and preparation just to step outside. Relatively simple things that will just make the lives of the humans easier and safer like:

"I left a wrench outside on top of generator 6. Please go retrieve it"

"We think there's a pressure leak in module 3 near the middle of the outside wall. Go inspect that area for any faults and take this sniffer and see if it detects anything in that area."

"Solar array 11 has accumulated some dust [from a storm on mars, or a nearby rocket landing on the moon]. Take this sweeper and clean off whatever you can reach."

"Take the trash out"