Humanoid Robots for Moon and Mars

[Coastal Ron]

While wheels are indeed efficient on flat surfaces, equipping robots with legs provides them with greater flexibility and adaptability. Legs allow robots to navigate a variety of terrains and obstacles that wheels might struggle with. This means they can be assigned to a wider range of tasks beyond just moving across a flat factory floor, making them more versatile in different work environments.

Humans and robotics in a factory tend to be focused on a limited number of tasks that they do, and pretty much none of them would require a change in elevation such as steps or ramps.

There are applications where moving across uneven terrain is required, but outside of the military I think the market for that is pretty low right now here on Earth, but the lack of infrastructure on the Moon and Mars will require robotic systems that can move around on their own.

Robotic systems with legs could have a higher "crash" rate than systems with wheels, so that needs to be taken into account. And what will the jobs be for these robotic systems? If the job is unloading cargo, and maintaining a warehouse, then wheeled versions will probably be predominant. If the job is to explore uneven terrain, then robotic systems with legs would probably be dominant at the exploration site, but would use wheeled vehicles to get to the exploration site.

One thing to keep in mind for both the Moon and Mars is that even here on Earth independent robotic systems can't do much for very long before requiring a recharge, and on the Moon and Mars that would likely be worse because of the need for heaters. So operating close to charging stations will be mandatory, and part of the infrastructure buildup will be in putting in charging stations.

[InterestedEngineer]

While wheels are indeed efficient on flat surfaces, equipping robots with legs provides them with greater flexibility and adaptability. Legs allow robots to navigate a variety of terrains and obstacles that wheels might struggle with. This means they can be assigned to a wider range of tasks beyond just moving across a flat factory floor, making them more versatile in different work environments.

Humans and robotics in a factory tend to be focused on a limited number of tasks that they do, and pretty much none of them would require a change in elevation such as steps or ramps.

You must not shop at box stores much.  Ladders are still in common use at warehouses and warehouse stores.  See attached.

If you watch the entire video, you will note they discuss the tradeoff between general purpose and dedicated robots.  General purpose (aka humanoid), have many uses, including for those it is hard to plan for all contingencies in advance.

I suspect on Mars there are numerous contingencies that you can't plan for in advance.  Such as for a simple example the lift to a Starship being broken and having to break out a ladder to get up and down to the cargo compartment.

[daedalus1]

I still don't get why they need legs to move on a flat factory shop floor, when wheels would be simpler and more efficient.

Robot are being designed to mimic us as much as possible, because we build our world around  us. Put wheels on us and you will soon realise how restrictive that is.

[Cheapchips]

One thing to keep in mind for both the Moon and Mars is that even here on Earth independent robotic systems can't do much for very long before requiring a recharge, and on the Moon and Mars that would likely be worse because of the need for heaters. So operating close to charging stations will be mandatory, and part of the infrastructure buildup will be in putting in charging stations.

Slap a minimal space suit on them when they work outside. It sorts out the heating and cooling needs without requiring specialist space versions of what are mass produced items (assuming viability of the whole humanoid robot sector).  1X's Neo already wears a onesie. It's not much of a stretch.

Battery life isn't really much of an issue when you have a small fleet of 'bots and navigating to a charge station is probably the simplest task you can ask of them.

[sanman]

The recent demonstration of Tesla's Optimus robot catching a ball looked like yet another new milestone in achieving human-like movement. One thing to note was that the robot was tele-operated by a human for this demo. I guess the robot's realtime AI processing isn't on par with human abilities yet.

[Coastal Ron]

One thing to keep in mind for both the Moon and Mars is that even here on Earth independent robotic systems can't do much for very long before requiring a recharge, and on the Moon and Mars that would likely be worse because of the need for heaters. So operating close to charging stations will be mandatory, and part of the infrastructure buildup will be in putting in charging stations.

Slap a minimal space suit on them when they work outside.

If the suit is pressurized then you have the same problem that humans do, which is fighting the inflated bag you're in to move any joint. Not really the right solution since that increasing power needs.

It sorts out the heating and cooling needs without requiring specialist space versions of what are mass produced items (assuming viability of the whole humanoid robot sector).  1X's Neo already wears a onesie. It's not much of a stretch.

Putting insulation on robots is likely the best solution, but you still have heat loss.

Battery life isn't really much of an issue when you have a small fleet of 'bots and navigating to a charge station is probably the simplest task you can ask of them.

I am on my 2nd electric car, and I'm quite aware of the charging infrastructure that is needed for large batteries. In order to have charging stations you need to ship all of them there, then ship the copper cabling between them and the battery station (which you have to ship there too), and then connect the batteries up to the power generation system - which nobody knows for sure what that will be.

Waving hands and saying it is simple is ignoring reality...

[Coastal Ron]

While wheels are indeed efficient on flat surfaces, equipping robots with legs provides them with greater flexibility and adaptability. Legs allow robots to navigate a variety of terrains and obstacles that wheels might struggle with. This means they can be assigned to a wider range of tasks beyond just moving across a flat factory floor, making them more versatile in different work environments.

Humans and robotics in a factory tend to be focused on a limited number of tasks that they do, and pretty much none of them would require a change in elevation such as steps or ramps.

You must not shop at box stores much.

I do, but notice I was specifically talking about factories, not consumer goods.

Ladders are still in common use at warehouses and warehouse stores.

If you'll notice in warehouses, they have LOTS of infrastructure that is required before you start storing inventory. That won't be the case on Mars. Mars will start out with cargo being stored out in the open on their shipping pallets until needed, then likely slowly migrate into storing pallets in semi-protected shelters until pressurized shelters can be built. Even then, there will be inventory that will just stay "outside" until it is needed.

So if that happens, then wheeled vehicles will be able to manage most of the inventory management. You could need humanoid type robots for breaking down pallets and retrieving smaller items, but for the big stuff you'll use the Mars equivalent of forklifts.

If you watch the entire video, you will note they discuss the tradeoff between general purpose and dedicated robots.  General purpose (aka humanoid), have many uses, including for those it is hard to plan for all contingencies in advance.

I didn't watch the video, but I've tracked the robotic sector since college, when I built my first robotic digit out of balsa wood and antenna wire. And what is really the limitation these days is that general robot hands are not yet very good for warehouse work. They lack sensors to know how they are gripping things, and vision systems are still working out how to grab things. I have a friend making software to control robotic systems for small manufactures to use for small lot production, so we talk about this type of stuff quite often. If anything, mobility (i.e. walking) is progressing faster than manipulators.

I suspect on Mars there are numerous contingencies that you can't plan for in advance.  Such as for a simple example the lift to a Starship being broken and having to break out a ladder to get up and down to the cargo compartment.

Or just use a crane...

[Twark_Main]

glances at title

Since "droid" (short for anDROID) is trademarked by Lucasfilm-cum-Disney, does this mean in the real world we'll use the word noid (humaNOID)? 



https://en.wikipedia.org/wiki/The_Noid#Chamblee_hostage_incident

[Cheapchips]

One thing to keep in mind for both the Moon and Mars is that even here on Earth independent robotic systems can't do much for very long before requiring a recharge, and on the Moon and Mars that would likely be worse because of the need for heaters. So operating close to charging stations will be mandatory, and part of the infrastructure buildup will be in putting in charging stations.

Slap a minimal space suit on them when they work outside.

If the suit is pressurized then you have the same problem that humans do, which is fighting the inflated bag you're in to move any joint. Not really the right solution since that increasing power needs.

It sorts out the heating and cooling needs without requiring specialist space versions of what are mass produced items (assuming viability of the whole humanoid robot sector).  1X's Neo already wears a onesie. It's not much of a stretch.

Putting insulation on robots is likely the best solution, but you still have heat loss.

Battery life isn't really much of an issue when you have a small fleet of 'bots and navigating to a charge station is probably the simplest task you can ask of them.

I am on my 2nd electric car, and I'm quite aware of the charging infrastructure that is needed for large batteries. In order to have charging stations you need to ship all of them there, then ship the copper cabling between them and the battery station (which you have to ship there too), and then connect the batteries up to the power generation system - which nobody knows for sure what that will be.

Waving hands and saying it is simple is ignoring reality...

By spacesuit I was envisaging whatever's the simplest effective garment to enable a consistent temperature for the actuators. As you say, insulation alone probably doesn't cut it. It doesn't need pressurised to the levels of a human suit with an OČ atmosphere. The power /performance impact doesn't have to be in the same ballpark.

On charging, I own an EV too. 85% of our kWs come from plugging it in on our drive.  The bots return home, be it a ship or base.  A Moon/Mars robot charging network isn't required.

For a remote job, drop the bots off with a big battery crate.  They can hook up a solar array to that if it's a scenario where that would extend the on site time.

The recent demonstration of Tesla's Optimus robot catching a ball looked like yet another new milestone in achieving human-like movement. One thing to note was that the robot was tele-operated by a human for this demo. I guess the robot's realtime AI processing isn't on par with human abilities yet.

That video is quite a flex for low latency remote operation. So long as humans are in situ, the capability to put human hands on a job without requiring an EVA is a really powerful tool.

[InterestedEngineer]

Mars will start out with cargo being stored out in the open on their shipping pallets until needed, then likely slowly migrate into storing pallets in semi-protected shelters until pressurized shelters can be built. Even then, there will be inventory that will just stay "outside" until it is needed.

You have noticed the rough ground on Mars photographs, right?

[Greg Hullender]

A new article in The Economist, Robots can learn new actions faster thanks to AI techniques, describes the use of "Large Behavior Models" (LBM) to use deep-net AI to train robots. This is a beautiful use of AI, most more practical that most of what I hear people talking about.

An aspect not mentioned in the article is that this technique could also be used to train very non-human robots. E.g. a six or eight-legged crawler might do a much better job on rugged terrain than a two-legged humanoid robot. You could always give it a humanoid torso (or two) if it needs to have hands. The advantage of using an LBM is that the training effort drops from years to hours. That frees you to try all kinds of different architectures that wouldn't have been worth taking a chance on before.

[Coastal Ron]

Mars will start out with cargo being stored out in the open on their shipping pallets until needed, then likely slowly migrate into storing pallets in semi-protected shelters until pressurized shelters can be built. Even then, there will be inventory that will just stay "outside" until it is needed.

You have noticed the rough ground on Mars photographs, right?

You're kidding, right?

One of the first pieces of equipment off of the first Starship to land will be some sort of wheeled vehicle with a blade attachment. Then they will grade the area for walking and driving paths, and the storage areas. Take the blade attachment off and the vehicle can go about the other duties it has.

Even with smart articulate robots you don't want to have to be stepping over rocks and uneven ground around your base, so grading needs to be a very early priority. Heck, you don't want humans stumbling over rocks and uneven surfaces either, so grading is not only needed for storing your inventory, but for safety reasons too.

But for applications where grading won't/can't be done, just add some fold out legs to the pallets... 

[darkenfast]

glances at title

Since "droid" (short for anDROID) is trademarked by Lucasfilm-cum-Disney, does this mean in the real world we'll use the word noid (humaNOID)? 



https://en.wikipedia.org/wiki/The_Noid#Chamblee_hostage_incident

Nah, just call them "'roids", because dealing with them will be a pain in the ... .

[redneck]

Mars will start out with cargo being stored out in the open on their shipping pallets until needed, then likely slowly migrate into storing pallets in semi-protected shelters until pressurized shelters can be built. Even then, there will be inventory that will just stay "outside" until it is needed.

You have noticed the rough ground on Mars photographs, right?

You're kidding, right?

One of the first pieces of equipment off of the first Starship to land will be some sort of wheeled vehicle with a blade attachment. Then they will grade the area for walking and driving paths, and the storage areas. Take the blade attachment off and the vehicle can go about the other duties it has.

Even with smart articulate robots you don't want to have to be stepping over rocks and uneven ground around your base, so grading needs to be a very early priority. Heck, you don't want humans stumbling over rocks and uneven surfaces either, so grading is not only needed for storing your inventory, but for safety reasons too.

But for applications where grading won't/can't be done, just add some fold out legs to the pallets... 

I've been working construction for a while. Storing materials in rough terrain is done all the time on some sites. We prefer flat ground, but that is not always practical. Wheels and tracks seem to go pretty much anywhere we need including 30 degree slopes with operators in the cab. Probably 45-60 degree slopes with remote operation of machines designed for it. I'm not sold on legs being the answer though I could be persuaded.

Some of our machines will take a half dozen attachments from blades to forks to buckets to work platforms. It seems possible to me that a fairly mundane solution will be used off world based on current machinery.

[InterestedEngineer]

Mars will start out with cargo being stored out in the open on their shipping pallets until needed, then likely slowly migrate into storing pallets in semi-protected shelters until pressurized shelters can be built. Even then, there will be inventory that will just stay "outside" until it is needed.

You have noticed the rough ground on Mars photographs, right?

You're kidding, right?

One of the first pieces of equipment off of the first Starship to land will be some sort of wheeled vehicle with a blade attachment. Then they will grade the area for walking and driving paths, and the storage areas. Take the blade attachment off and the vehicle can go about the other duties it has.

Even with smart articulate robots you don't want to have to be stepping over rocks and uneven ground around your base, so grading needs to be a very early priority. Heck, you don't want humans stumbling over rocks and uneven surfaces either, so grading is not only needed for storing your inventory, but for safety reasons too.

But for applications where grading won't/can't be done, just add some fold out legs to the pallets... 

I've been working construction for a while. Storing materials in rough terrain is done all the time on some sites. We prefer flat ground, but that is not always practical. Wheels and tracks seem to go pretty much anywhere we need including 30 degree slopes with operators in the cab. Probably 45-60 degree slopes with remote operation of machines designed for it. I'm not sold on legs being the answer though I could be persuaded.

Some of our machines will take a half dozen attachments from blades to forks to buckets to work platforms. It seems possible to me that a fairly mundane solution will be used off world based on current machinery.

Who is going to operate the machinery?  You going to build AI and all it's loooong training into it, while still having manual controls for when humans show up?

or use the AI for a humanoid robot, whose baseline is already trained, and use the robot to operate the machinery?

Who is going to reach in the side and fix a cable that has broken?  Not a tracked vehicle.

Who is going to set up and test the airlocks on the buildings?  A tracked vehicle?

[redneck]

Mars will start out with cargo being stored out in the open on their shipping pallets until needed, then likely slowly migrate into storing pallets in semi-protected shelters until pressurized shelters can be built. Even then, there will be inventory that will just stay "outside" until it is needed.

You have noticed the rough ground on Mars photographs, right?

You're kidding, right?

One of the first pieces of equipment off of the first Starship to land will be some sort of wheeled vehicle with a blade attachment. Then they will grade the area for walking and driving paths, and the storage areas. Take the blade attachment off and the vehicle can go about the other duties it has.

Even with smart articulate robots you don't want to have to be stepping over rocks and uneven ground around your base, so grading needs to be a very early priority. Heck, you don't want humans stumbling over rocks and uneven surfaces either, so grading is not only needed for storing your inventory, but for safety reasons too.

But for applications where grading won't/can't be done, just add some fold out legs to the pallets... 

I've been working construction for a while. Storing materials in rough terrain is done all the time on some sites. We prefer flat ground, but that is not always practical. Wheels and tracks seem to go pretty much anywhere we need including 30 degree slopes with operators in the cab. Probably 45-60 degree slopes with remote operation of machines designed for it. I'm not sold on legs being the answer though I could be persuaded.

Some of our machines will take a half dozen attachments from blades to forks to buckets to work platforms. It seems possible to me that a fairly mundane solution will be used off world based on current machinery.

Who is going to operate the machinery?  You going to build AI and all it's loooong training into it, while still having manual controls for when humans show up?

or use the AI for a humanoid robot, whose baseline is already trained, and use the robot to operate the machinery?

Who is going to reach in the side and fix a cable that has broken?  Not a tracked vehicle.

Who is going to set up and test the airlocks on the buildings?  A tracked vehicle?

Humanoid does not equal smart enough to do the job. Legs and arms don't create the intelligence to do the tasks. Just as a self driving car does not have a robot in the seat, and autopilots don't either, it will be the control system that gets the job done. There are already remote systems for some construction and mining equipment. So far, humans are more versatile, though autonomous forklifts are a thing as are drone deliveries.

Bottom line is that putting legs and arms on your robot is insufficient for creating control systems.

[sanman]

Yet another wheeled ambulatory drone -- seems like a next step improvement over the quadrupedal robodog idea.
Could it be useful for surface exploration of Moon, Mars, etc? What are the limitations?

[Greg Hullender]

The picture illustrates the idea very well. The human is riding the robot like a horse. Of course, in reality, you'd want the human to control it remotely--not actually ride it in person. The AI trains the robot to know how to move, but it doesn't give it any intelligence as to when or why to move. The human provides the volition, tells it which way to go, etc. It might report that it can't go that way, but that's it.

[sanman]

The picture illustrates the idea very well. The human is riding the robot like a horse. Of course, in reality, you'd want the human to control it remotely--not actually ride it in person. The AI trains the robot to know how to move, but it doesn't give it any intelligence as to when or why to move. The human provides the volition, tells it which way to go, etc. It might report that it can't go that way, but that's it.

I'm not sure if you watched the video itself, but it mostly shows the machine doing that.

I think applications like these show the need for high-performance edge computing. The more processing power you can give to your robot to locally process the terrain handling, then the more capability there can be to traverse terrain that would otherwise have previously been untraversable.

[sanman]

NASA is using robot dogs to explore caves in the Pacific Northwest