Musings on telerobotics

[docm]

I've seen a lot of posts on the net full of hand-wringing over the idea of the "deep space" missions being 'no touch', meaning most would be no-landing observation tours.

Thought about it for a while and came to the conclusion that many of these missions (Mercury, Venus, large asteroids and planetoids etc.) might be best done using a mix of humans in orbit or in parallel trajectories and a new generation of rovers using 'telerobotics' - humans controlling robot missions from a closer range than ever possible before.

I know NASA had been working on a human-robot interface for ISS, but I think that paradigm needs to go further.

Humans do have this 'thing' about reaching out and touching a new environment under control of our stereoscopic peepers. Heft the rock, fine motor control of the geology tools, close inspection etc. are some of the advantages of humans on the rock.

Using telerobotics on the surface with humans within light-milliseconds would allow for realtime manipulation, which should allow more science to be done in a much shorter time.

Sad to say the bucks aren’t there for the personal touch. Maybe some day, or on the other hand maybe telerobots could be built with sensory feedback and HD stereo vision to make the issue near to moot.

Recent developments in robotic prostheses by DARPA & others plus advances in virtual reality are allowing for very fine control of robotic arms and hands, in addition to information like pressure, texture, temperature, touch etc. being passed along to a human controller remotely. Same for remote HD stereoscopic viewing.

Last I heard DARPA's hand had 25 ways of freedom, only a few short of a real humans and it had sensory feedback. Arms are advancing so fast it's hard to keep track of them. Stereo HD remote vision systems are already there.

Perfect for rovers with at least one human-like arm controlled by a human in close proximity to the space body under study.

Savings would come in reduced infrastructure: no hab on the ground, no lander/return vehicle/manned rover/environmental gear etc.

 

[3488]

Hi docm,

I like that. I cannot see why a test mission could not be sent to a scientifically interesting location on the near side of the Moon, say the mountains in the centre of Copernicus Crater, or the Ina D Caldera, etc.

The Moon is only 1.5 light seconds away, so assuming a robot responds immediately we would get a reply in only 3 seconds.

This test mission would validate telerobotics & would pave the way for as you say Mercury, Venus, Mars, 1 Ceres, 2 Pallas, 4 Vesta, NEAs, etc with a human crew in orbit or close by, lets say Phobos & or Deimos in the case of Mars.

I like it & think that this shuold be properly funded & actioned.

Great subject.

Andrew Brown.

 

[neuvik]

Cool idea. I'd assume however that if a rover has to transmit data concerning the ammount of pressure its fingers apply to an object would increase power consumption and wieght.

But I think your right on the money, just send a crew in to orbit, and use remote control units to do the planetary surface science. I'd bet that saves money, as they won't need a lander capable of returning to orbit, and takes away more concerns from a safety point. What if you trip, space suit life support malfunctions, or weather change (assuming the celestial body does infact have an atmosphere).

 

[docm]

Cool idea. I'd assume however that if a rover has to transmit data concerning the amount of pressure its fingers apply to an object would increase power consumption and weight.

Not as much as you'd think.

The sensors used in the DARPA and other hands amount to a thick finger cot with serial digital + power wiring, and their use has been tested in amputees going back at least 2 years I know of. Some patients have adapted so well they can differentiate between different grades of sandpaper and pick up foam cups without crushing them, which the hands are very capable of doing....they're very powerful actually.

I think USB 2 has been used by some groups, but USB 3 is very close to market, next year in fact, and it is vastly faster (3.2 GBit/s after overhead), and remember USB not only provides digital serial data but power so no additional circuit for that. One small diameter wire for data and power.

Use a single cable as the 'nerve' down the robotic arm to a hub then split it to the individual sensors and/or controls. Remember: each primary USB circuit can be split among up among 128 devices, so a single hub in the chassis could be split into 64 channels each for a pair of arms. That or give each arm its own circuit for a full compliment of 128 channels per arm.

But I think your right on the money, just send a crew in to orbit, and use remote control units to do the planetary surface science.

I thought it sounded reasonable, but needed feedback. Thanks.

 

[neuvik]

No I'm sure its not as much as I think. I was just remarking some possible negative aspecs for the sake of argument.
I think the touch feedback would be a good idea in some places. But overall I doubt it would be that important.

For the use of instruments and equipment you could just have a forklift like slide to wedge in to receptcales in science devices and tools. I use fork lift as an example, I'd imagine the assembly for Bobcats, or like small construction vehicles would be far more applicable. If your not familure with Bobcats and similar construction equipment, they have a plate assembly with several hydraulically activated teeth. The Bobcat just need to mate the plate with the backplate of a tool and activate the teeth. This allows one vehical to do many differant tasks. (I'm sure you knew this but for everyone esle.)

Also there is pretty much a machine sense you get when operating a vehical for a long enough time. For instance a few years back when I did a logging stint in Norther Califronia, we had to be very careful working around creek and rivers. I was using a Hyundai LC290 and I swear I could probably pick a baby from its cradle with it. I moved the logging trailers, and yarders around some tight bends, along with a myrid of semi delicate things; even the occasional person who wanted a lift out of a trench or bank.


Back to the electronics though, USB 3 uses more power than USB 2, just a small note. I agree its nice to have the communication run along with power. But for imaging that makes a problem with harmonics. USB 3 is addressing that, but I have no idea about the details. USB 3 also needs a larger bank of address lines for the controller. Its gotta accomated the identification of the devices along with the information they are sending / recieving. At least heating won't be a problem.

 

[docm]

My emphasis on sensory feedback & robotic arms has a benefit beyond tactile - easier training. It takes a lot of time to become proficient with remote manipulators. I know...been there, done that. Everyone knows how to work their arm, and the body learns very rapidly how to transfer ones mental presence to a robot. That study came out 6 months ago and it was stunning how fast the participants adapted - sometimes hours. They actually started thinking of the robot body as their own, not thinking at all about the user interface.

A USB 3 host controller chip is about the size of a thumbnail and generates about enough heat to keep a cockroach warm.

NEC's

As for video on USB - I have this little USB 2 box on my desk that picks up 720p HD broadcasts and feeds them into a computer in real time for display. Great pics, works like a charm, no noise with a good S/N and cost about $150. I understand the NEC USB 3 host controller can transfer about 25 gigs of video per minute.

My point is that it's very possible multimedia companies are already making solutions to the problem you iterated. Even probable.

Hell...2 years ago if someone told me that someone would be making a consumer camcorder that recorded to memory cards and could be used underwater for $300+ I'd have called them nuts. Two came out within the last 9 months.

 

[neuvik]

And I've ran out of negative arguments for the rover... But we havn't adressed the ships ferrying the crew!

It would be cool to get some joint studies to develope something like this, and as Andrew suggested, test it on the moon. I'd think DARPA would get as much useful infromation from such an endevour working with NASA than alone. Such a device we would find usefull, other than artificial limbs and the obvious rover, would be a unit to clean up HAZMAT sites, mine fields, and many more hostile enviroment duties.

 

[3488]

Already a telerobot is used in Israel, etc whenever a suspect package is seen underneath a bus. A remote controlled robot is sent out equipped with cameras & arms & can either defuse it or detonate it safely.

We alrady have the technology to make this doable for planetary exploration. We lack the infrastructure.

Andrew Brown.

 

[docm]

We alrady have the technology to make this doable for planetary exploration. We lack the infrastructure.

Andrew Brown.

More lacking is the will and funding. The infrastructure can be built quickly once you have those two items. Just don't turn it over to formerly corporate tortoises. Do give it to the hungry young Turks with guidance from a 'bot crazy engineer who won't quit.

 

[kelvinzero]

I have always been enthusiastic about this direction. At some point a teleoperated experience will exceed what is possible though a spacesuit.

On a slightly related topic I have been thinking our senses are not very well designed for space anyway.
Take the moon for example. To our vision it has painful contrasts with little colour and it is very difficult to judge distances.

It would be better if we could use some sort of augmented reality: apply a digital distance haze that tints the sky and distant mountains blue, for comfort on the eyes and better distance perception. Reduce contrasts of intensity with HDR image processing so you can see into shadows even when next to objects in direct sunlight. Increase contrasts of colour so variations in regolith composition are visible at a glance.

 

[docm]

Not a prosthetic hand/arm, but this is how far realtime robotic manipulation + artificial sight has come. There is no way a telerobot using these strategies couldn't keep up with its remote operator, and then some.

[youtube]http://www.youtube.com/watch?v=OTdpr89Dk0w[/youtube]

 

[3488]

Thanks docm, fascinating indeed. I agree these can keep up with their operators, that's for certain.

Below is a simple demonstration of how a simple control panel on the internet could be used for telerobotics. Certainly this will evolve to be used on far more capable hardware.
[youtube]http://www.youtube.com/watch?v=hVqx2fnYi0A[/youtube]

Andrew Brown.

 

[MeteorWayne]

Doing a bit of housecleaning on Missions and Launches today. This thread isn't really either, so I'll move it to Space Business and Technology where it's a better fit.

 

[neutrino78x]

Ah yes, the age old issue of whether to send humans for purely science missions! lol.

I am on the side of using robots for pure science, to the extent that it is possible.

I think the ultimate issue is actually the speed of light. Once the mission is far enough from the Earth, all it can do with human operators is take pictures and other observations which don't require real time manipulation.

For example, you can fly around the Jovian system all you want, take pictures, use radar, a magnetometer, etc., but what if you wanted to put a probe in Europa and look around inside the water ocean that is suspected there? Like the way that they look at the Titanic on the bottom of the sea? Granted, the probe could simply go underwater and maybe do some predetermined motions, but what if there really is life down there, like some kind of fish or something, and it goes by the camera? You can't follow after it, because by the time you have seen it happen on Earth, it was hours ago. So, the best scenario at Jupiter would be to either build a base at Callisto, or have a nuclear powered ship with artificial gravity orbiting. Then the speed of light delay is negligible, and you can move the submarine rover around in real time.

--Brian

 

[JonClarke]

Teleoperation is all very well but it comes at a cost in terms of power, volume, mass, and communications requirements. DEXTRE for example masses 1.5 tonnes and uses 1.2 kW. To be used at their full potential they are also manpower intensive, and are still nowhere near as capable in terms of dexterity, flexibility, or sensor capability as an astronaut, even wearing today's suits.

So before we send fleets of teleoperated machines let's see what they cost and what they really are capable of. There are a lot of expectations about teleoperation (as with "robots" generally) that are generated more by fantasy than reality.

Jon