Author Topic: General purpose worker droids  (Read 5816 times)

Offline CameronD

  • Full Member
  • ****
  • Posts: 1403
  • Melbourne, Australia
    • Norton Consultants
  • Liked: 450
  • Likes Given: 334
Re: General purpose worker droids
« Reply #20 on: 03/08/2018 01:59 AM »
What happened with Robonaut 2? I was expecting it would be supporting dangerous EVA’s by now.
Project M (in centaur version) would have been so inspirational :)

Lessee.. spend millions of $$$ on a one-way ride to the Moon, where it tosses a rock and then stands there looking at Earth?!?  There has to be something better to do than that.

No Mission, no Money.  ;)
« Last Edit: 03/08/2018 02:01 AM by CameronD »
With sufficient thrust, pigs fly just fine - however, this is not necessarily a good idea. It is hard to be sure where they are
going to land, and it could be dangerous sitting under them as they fly overhead.

Offline Asteroza

  • Full Member
  • ****
  • Posts: 626
  • Liked: 81
  • Likes Given: 2
Re: General purpose worker droids
« Reply #21 on: 03/08/2018 10:05 PM »
Robonaut 2 is riding a Dragon back to earth soon. Haven't heard of it going back up either...

Offline Aussie_Space_Nut

  • Full Member
  • *
  • Posts: 196
  • South Australia
  • Liked: 67
  • Likes Given: 205
Re: General purpose worker droids
« Reply #22 on: 03/10/2018 01:38 AM »
I'm with QuantumG on this one.

I am a CNC Machinist/Programmer/CAD Draughtsperson. I have had serious conversations with very intelligent people, way more intelligent than me, who honestly believe this is easy, one of whom insisted he could write an algorithm to do my CNC job of programming. As I said to him at the time if it was that easy it would have been done a long time ago because of the enourmous cash cow it would become.

The "robots" of today always have a human near by teaching them or programming them to do repetative tasks and yes rescuing them when required. You don't often see these humans on the youtube clips.

We won't see the robots of our sci-fi movies until someone seriously cracks AI. If that should happen though all bets are off. Intelligence, the ability to problem solve using all of the resources you have at hand, that is the key to our sci-fi robot dreams. Just not seeing it ON THE SAME LEVEL as a reasonably competant human at this point.
« Last Edit: 03/10/2018 01:39 AM by Aussie_Space_Nut »

Offline Aussie_Space_Nut

  • Full Member
  • *
  • Posts: 196
  • South Australia
  • Liked: 67
  • Likes Given: 205
Re: General purpose worker droids
« Reply #23 on: 03/10/2018 02:30 AM »
I was amused to see the article below after just posting in this thread.

Key statement that got my attention, "starts the burger assembly process for staff."

http://www.abc.net.au/news/2018-03-10/flippy-the-burger-flipping-robot-has-been-taken-offline/9535192

Offline Symmetry

Re: General purpose worker droids
« Reply #24 on: 03/13/2018 03:34 PM »
I work as a professional roboticist trying to automate new parts of the logistics pipeline.  We're not going to see fully autonomous general purpose robots soon (within 20 years).  What we could see would be, e.g., more specialized robots for things like bolting together struts.  The ISS has the very cool Canadarm2 on board and you can make things more autonomous and flexible than that.  There's a wide range between an arm that has to be continuously operated by a human and a fully autonomous system.  You can have human supervision or systems that only require a human to step in when the robot recognizes a circumstance it is unable or less able to handle on its own.  LEO doesn't have much speed of light delay so there's lots of scope for ground supervision and less need for autonomy than robots on Mars have.

Also, I don't see why you would want a bipedal robot in microgravity and I'm pretty skeptical of them even in environments built for humans.

If you want to see what autonomous robots are capable of right now look for videos from the DARPA Robotics Challenge on YouTube.

Offline whitelancer64

Re: General purpose worker droids
« Reply #25 on: 03/13/2018 04:16 PM »
What happened with Robonaut 2? I was expecting it would be supporting dangerous EVA’s by now.
*snip*

Robonaut 2 developed a fault in its power system about 2 years ago. They were unable to resolve the issue on the ISS, so it's being shipped back to Earth for repairs. As Asteroza said, it's uncertain whether it will be going back to the ISS, but I would guess it probably will since they will want to do more in-space testing with it.
"One bit of advice: it is important to view knowledge as sort of a semantic tree -- make sure you understand the fundamental principles, ie the trunk and big branches, before you get into the leaves/details or there is nothing for them to hang on to." - Elon Musk
"There are lies, damned lies, and launch schedules." - Larry J

Offline su27k

  • Full Member
  • ****
  • Posts: 1112
  • Liked: 811
  • Likes Given: 75
Re: General purpose worker droids
« Reply #26 on: 03/13/2018 05:05 PM »
Dozens of self-driving cars have been sharing the roads with us for years.  They're expensive, but they are very good at autonomously navigating a vast environment that has not been specially prepared for them.

Really? What cars are those? "Self-driving cars" are currently more accurately described as driver assistance.

Roads are specially prepared environments... mostly for human drivers.

There's no general AI in these devices. That's my point.

Google is already testing Level 4 autonomous cars on public road: http://www.thedrive.com/tech/15848/waymo-is-already-running-cars-with-no-one-behind-the-wheel, that's no driver assistance.

Of course there's no "general" AI in these, it's still a somewhat narrow AI application, but the progress made is impressive, and similar approach may be applicable to a huge segment of what a human worker do everyday.

Offline DarkenedOne

  • Full Member
  • ****
  • Posts: 954
  • Liked: 57
  • Likes Given: 8
Re: General purpose worker droids
« Reply #27 on: 03/13/2018 05:48 PM »
Dozens of self-driving cars have been sharing the roads with us for years.  They're expensive, but they are very good at autonomously navigating a vast environment that has not been specially prepared for them.

Really? What cars are those? "Self-driving cars" are currently more accurately described as driver assistance.

Roads are specially prepared environments... mostly for human drivers.

There's no general AI in these devices. That's my point.

Google is already testing Level 4 autonomous cars on public road: http://www.thedrive.com/tech/15848/waymo-is-already-running-cars-with-no-one-behind-the-wheel, that's no driver assistance.

Of course there's no "general" AI in these, it's still a somewhat narrow AI application, but the progress made is impressive, and similar approach may be applicable to a huge segment of what a human worker do everyday.

As far as robotics is concerned getting a car to drive itself is not difficult.  Autonomous mining vehicles and similar systems have existed for a while now.  Really all you need is precise maps and good localization.  What's difficult with autonomous cars are the people.  They cannot mapped nor are they predictable. 
« Last Edit: 03/13/2018 05:50 PM by DarkenedOne »

Offline QuantumG

  • Senior Member
  • *****
  • Posts: 8638
  • Australia
  • Liked: 3548
  • Likes Given: 840
Re: General purpose worker droids
« Reply #28 on: 03/13/2018 08:03 PM »
Google is already testing Level 4 autonomous cars on public road

Yeah, so? I really don't understand why people have difficulty understanding that research isn't product.

I hear those things are awfully loud. It glides as softly as a cloud. What's it called? Monowhale!

Offline Darkseraph

  • Full Member
  • ****
  • Posts: 577
  • Liked: 271
  • Likes Given: 119
Re: General purpose worker droids
« Reply #29 on: 03/13/2018 09:00 PM »
Best that can be hoped for in the near future are specific purpose robots with human babysitters to fix them, troubleshoot and intervene for non-repetitive tasks that require either human dexterity or human problem solving. Which will actually be probably good enough for most of what we would like to do in space. They will probably look like Canadarm on wheels (or on a satelite bus) with a few interchangeable end effectors stored on it.



Work by Made in Space, Tethers Unlimited, SSL and Orbital ATK for in space servicing and assembly broadly looks like what we can expect in a decade or two. 

For the purpose of building megastructures, probably some big long truss structure lined with arrays of industrial robotic arms that do preprogrammed repetitive tasks with a dozen or so human workers stationed in a small pressurized section that monitor and intervene in the event of a problem. Even that is all far in the future though. 

For reference something not too different from this


It's worth mentioning that being able to send actual human workers to space for far less than it currently costs will have a huge impact on the kinds of activities that are possible there. The current costs are tremendously high and so only six people are up there at any one moment and the ISS is underutilized. Dozens or even hundreds of people in space in combination with automated systems will allow amazing things to be achieved that are unthinkable today.
« Last Edit: 03/13/2018 09:10 PM by Darkseraph »
"For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled." R.P.Feynman

Offline su27k

  • Full Member
  • ****
  • Posts: 1112
  • Liked: 811
  • Likes Given: 75
Re: General purpose worker droids
« Reply #30 on: 03/14/2018 02:12 AM »
As far as robotics is concerned getting a car to drive itself is not difficult.  Autonomous mining vehicles and similar systems have existed for a while now.  Really all you need is precise maps and good localization.  What's difficult with autonomous cars are the people.  They cannot mapped nor are they predictable.

The current autonomous cars are not just following a precise map, they use sensors to get an idea of the surroundings and react according, that's the only way it can drive along with other humans. There's no difference between what the autonomous car is doing and what a human driver is doing, except the car has more sensors.

Google is already testing Level 4 autonomous cars on public road
Yeah, so? I really don't understand why people have difficulty understanding that research isn't product.

More like beta testing, last step in product development.

Offline ChrML

  • Member
  • Posts: 38
  • Liked: 13
  • Likes Given: 0

Offline alexterrell

  • Full Member
  • ****
  • Posts: 1624
  • Germany
  • Liked: 88
  • Likes Given: 62
Re: General purpose worker droids
« Reply #32 on: 03/17/2018 10:04 PM »
We watched science fiction with general purpose robots, and expected something like Sonny in A.I. to be cutting our lawns. Instead, we end up with little specialist drones that cut the lawn over a 1 hour period - and then go and recharge. They do this great, but aren't so good at making the tea.

So the trend has seen huge improvements in specialist robots, doing their task. But very little in general purpose robots.

Linked question- would we be closer to a general purpose remote control droid, with a human on Earth in an exo-skeleton, and a machine on the moon doing what he does?

Offline Bob Shaw

  • Full Member
  • ****
  • Posts: 1063
  • Liked: 432
  • Likes Given: 376
Re: General purpose worker droids
« Reply #33 on: 03/17/2018 10:51 PM »

In that case, here's your General Purpose Worker Droid.  ;)

Sure, they're a bit bulky, require lots of prep and tend to be rather finicky to operate.. but they work well in swarms and are existing tech.



Also, they can be manufactured using unskilled labour.

Offline ChrML

  • Member
  • Posts: 38
  • Liked: 13
  • Likes Given: 0
Re: General purpose worker droids
« Reply #34 on: 03/20/2018 04:12 PM »
Linked question- would we be closer to a general purpose remote control droid, with a human on Earth in an exo-skeleton, and a machine on the moon doing what he does?
Possible on the moon due to only a couple seconds latency. Excavation machines for water could be controlled like that until general AI is solved.

Altough not feasible on Mars due to latency. In that case, the human would have to work from within a habitat on Mars.

Offline LMT

  • Lake Matthew Team
  • Full Member
  • ****
  • Posts: 650
    • Lake Matthew
  • Liked: 74
  • Likes Given: 0
Re: General purpose worker droids
« Reply #35 on: 03/21/2018 01:07 AM »
Linked question- would we be closer to a general purpose remote control droid, with a human on Earth in an exo-skeleton, and a machine on the moon doing what he does?
Possible on the moon due to only a couple seconds latency. Excavation machines for water could be controlled like that until general AI is solved.

Altough not feasible on Mars due to latency. In that case, the human would have to work from within a habitat on Mars.

Lunar latency does present a challenge to the VR aspect of robotics, as with an exoskeleton telerobotic system. 

--

One story: 

Last year a space roboticist described to me his team's recent attempt to simulate lunar VR telerobotics, with 2.6 sec latency.  Their testers, all experienced in VR controls, quickly found the lunar latency intolerable, even sickening.  Their minds just couldn't compensate for the latency.  For those testers, even a few minutes of lunar VR was unpleasant.  A full VR workday would be out of the question.  Semi-autonomous robotics would be needed for any time-consuming tasks.

--

This might be a point in favor of a lunar base -- or the Deep Space Gateway.  Proper VR telerobotics would be feasible from a DSG in lunar orbit.

Likewise, yes, a Mars hab would support low-latency planet-wide VR telerobotics.  Robotics on Phobos and Deimos, too, with control from above or below.   A first base at Phobos, for example, could run VR telerobotics on Phobos, Mars and Deimos.
« Last Edit: 03/21/2018 01:17 AM by LMT »

Offline gin455res

  • Full Member
  • ***
  • Posts: 353
  • eny, sed, woz, shuga English spelling is messed up
  • bristol, uk
  • Liked: 11
  • Likes Given: 24
Re: General purpose worker droids
« Reply #36 on: 03/21/2018 06:50 AM »
Linked question- would we be closer to a general purpose remote control droid, with a human on Earth in an exo-skeleton, and a machine on the moon doing what he does?
Possible on the moon due to only a couple seconds latency. Excavation machines for water could be controlled like that until general AI is solved.

Altough not feasible on Mars due to latency. In that case, the human would have to work from within a habitat on Mars.

Lunar latency does present a challenge to the VR aspect of robotics, as with an exoskeleton telerobotic system. 

--

One story: 

Last year a space roboticist described to me his team's recent attempt to simulate lunar VR telerobotics, with 2.6 sec latency.  Their testers, all experienced in VR controls, quickly found the lunar latency intolerable, even sickening.  Their minds just couldn't compensate for the latency.  For those testers, even a few minutes of lunar VR was unpleasant.  A full VR workday would be out of the question.  Semi-autonomous robotics would be needed for any time-consuming tasks.

--

This might be a point in favor of a lunar base -- or the Deep Space Gateway.  Proper VR telerobotics would be feasible from a DSG in lunar orbit.

Likewise, yes, a Mars hab would support low-latency planet-wide VR telerobotics.  Robotics on Phobos and Deimos, too, with control from above or below.   A first base at Phobos, for example, could run VR telerobotics on Phobos, Mars and Deimos.


This is interesting. Do you have links to further research in this area?




I'm wondering if creating a god simulation like simcity or age-of-empires, but with a 2.6 sec latency and semi-autonomous agents might be worth investigating. The god sim might be more like a vr world and involve (spatial) task co-ordination of a higher complexity than a typical god-sim.


Was the experiment cited purely investigating the control of single machines? I see a more supervisory role (of teams of machines) being more fruitful.

Offline LMT

  • Lake Matthew Team
  • Full Member
  • ****
  • Posts: 650
    • Lake Matthew
  • Liked: 74
  • Likes Given: 0
Re: General purpose worker droids
« Reply #37 on: 03/21/2018 10:40 AM »
Was the experiment cited purely investigating the control of single machines?

Yes, immersive VR/video control of one dexterous robot with Earth/Moon latency.

links to further research in this area?

:)


Offline QuantumG

  • Senior Member
  • *****
  • Posts: 8638
  • Australia
  • Liked: 3548
  • Likes Given: 840
Re: General purpose worker droids
« Reply #38 on: 03/21/2018 09:43 PM »
Just showing the blooper reel is fun but I imagine they eventually learnt what everyone else learns in that situation: slow down, wait for the lag to catch up before taking your next action. The activities like cooking pancakes are learnable because they have "static" states. Playing ping pong, probably not.

On the vacuum cleaner front: I got a new one that gets caught on my curtains, mats, shoe laces, or any other "soft" junk that happens to be on the floor. After removing all those obstacles it manages to do a passable job of keeping the floors free of dust.

Running it daily trains me to keep things off the floor too. ::)

What makes this one better than the last one I had? It's better at mapping the environment. Does it have a clue what it is doing? Nope. The algorithms in this robot are about mapping the world, driving the area inside the map, and detecting when the brushes are turning or not. It doesn't know if the floor is clean or dirty. It doesn't spend more time on the messy bits than it does on the clean bits.

I expect future models* could have (better) cameras and do image processing (perhaps in the cloud) to evaluate how well it has cleaned - then they could start applying policy learning algorithms to produce better cleans. The concepts in those systems (probably sub-symbolic) would be about cleaning, not just getting around, and then you could reasonably start talking about the "intelligence" of these systems.

* As I understand it, iRobot has a mop robot that you can train manually to recognise when your floors are "clean" and when they need more than a once-over.



I hear those things are awfully loud. It glides as softly as a cloud. What's it called? Monowhale!

Offline Cinder

  • Full Member
  • ****
  • Posts: 626
  • Liked: 81
  • Likes Given: 383
Re: General purpose worker droids
« Reply #39 on: 03/21/2018 10:20 PM »
VR in laggy operation could also have some AR aids: visualization of your intended input.  E.G. the arm of the droid could be CG-rendered to reflect your commands in real time, while the droid's delayed video caught up.

For the purpose of building megastructures, probably some big long truss structure lined with arrays of industrial robotic arms that do preprogrammed repetitive tasks with a dozen or so human workers stationed in a small pressurized section that monitor and intervene in the event of a problem. Even that is all far in the future though. 


It's worth mentioning that being able to send actual human workers to space for far less than it currently costs will have a huge impact on the kinds of activities that are possible there. The current costs are tremendously high and so only six people are up there at any one moment and the ISS is underutilized. Dozens or even hundreds of people in space in combination with automated systems will allow amazing things to be achieved that are unthinkable today.
Bold/italics: why?  Is there a thread here or a book or online resource that describes your reasons accurately enough ? Ideally something recent enough to consider/refute projections based on all of the latest space industry developments.
What do you think of SpiderFab?

The unbolded part seems like too much of a dam trying to spill over the initial obstacles.
« Last Edit: 03/21/2018 10:30 PM by Cinder »
The pork must flow.

Tags: