It's only a matter of time now before robot nudity is a registerable sex offense...
Unfortunately, the announcement isn't that the robot has the dexterity to put on the clothes itself, but rather that its sensors and other functionality work through clothes, i.e., a hazmat suit.
Petman isn't doing any kind of active sensing (vision, radar/lidar, etc) in these videos. The clothing has nothing to obstruct other than movement, but it looks baggy enough such that it doesn't do this either.
I'm not sure if you're being cynical or sarcastic:
"...thanks to some DoD funding, it's using the robot to test suits like this in hazardous conditions, with an array of sensors on the robot itself able to detect chemicals leaking through the suit."
There's a clear purpose here. Humanoid robot testing of garments intended for hazardous duty. I doubt anyone at Boston Dynamics or the DoD are sitting around pondering how they're going to cloth the obscene masses of robot humanoids they plan to produce.
There is a video [1] of it walking on its own. I suspect they have the cables because it was leaning over, twisting, squatting, etc and they probably want to assist with balance.
They're probably fall-dampening cables with very light tension. These types of cables are like rate limiters for falling objects. It would prevent it from slamming down on the floor if balance was lost, but the tension is very light; just enough to take up slack. Because they're positioned only to the sides and the rear, their greatest effect would actually be to make balance more difficult for a bipedal robot.
The cables at the top are all for fall protection. The tether (power and control) goes through its ankle or hip area, depending on the video. The one video where you see multiple strands hanging down directly from the top are connected to a block and tackle pulley system. If you watch the one shot from behind you can see the hoist hook in the center of its shoulders up high. You can also see the tether at the hip area in the same shot.
The cables definitely look like training wheels. Even a very slight bit of tension could assist with balance. Of course that assumes the robot has some sort of kinesthetic sense.
I'd assume it's probably easier to iterate and test without worrying if your ridiculously expensive robot is going to fall on it's face. Even something small like friction with the clothes or binding of cloth in a servo/joint could cause it to fall, I suppose.
I'm sure they wanted their proof of concept out just to be the first robot with clothes, but just didn't want to risk damage.
The early versions of BigDog had support cables to protect it in case it fell as well. Eventually the stabilization hardware + software got good enough to where they weren't needed any more.
The cables are there to prevent the robot from damaging stuff during testing. If a human falls, it catches itself, or at worst breaks a wrist and spends a couple weeks healing. Unrestrained robots have far more spectacular failure modes; a falling humanoid typically lands on an outstretched limb and breaks it right in half, setting the project back months and costing huge amounts of money. Or it goes nuts but doesn't fall over immediately, instead flailing around on a short rampage that ends with the robot broken and everything nearby having been whacked by the limbed equivalent of a runaway golf cart.
Source: been there, done that, don't tell OSHA. :P
Don't worry, Anonymous will hack them and keep us safe.
Certainly, the power divide between the government and people (inside or outside of its borders) continues to grow. The scary thing to me (in a distopian future) is that robots won't hesitate to do as they are programed. Soldiers sent in to "subdue" their own population might at least think twice.
It has been interesting to watch Dr. Raibert's ideas grow. A long time ago there was an interesting debate going on between subsumption architecture that Dr. Brooks was pushing and Dr. Raiberts work (both at MIT). As a robot guy I've been fascinated by both of them, but Raibert's work was hampered by the fact that his creations were only stable when constantly moving and Brook's creations were always stable. The argument went that dynamic stability was so energy inefficient it would not be a good fit for doing locomotion.
The advantage of dynamic stability is of course that your robot is constantly in danger of falling down, that is its natural state, so trying to knock it over doesn't really change what its doing a whole lot.
I did not think I would see dynamic stability in a humanoid shape though (power off condition is laying horizontal :-). Watching it is pretty unnerving. I'm impressed they have it working.
22 comments
[ 2.8 ms ] story [ 68.0 ms ] threadUnfortunately, the announcement isn't that the robot has the dexterity to put on the clothes itself, but rather that its sensors and other functionality work through clothes, i.e., a hazmat suit.
Or a selling point....
"...thanks to some DoD funding, it's using the robot to test suits like this in hazardous conditions, with an array of sensors on the robot itself able to detect chemicals leaking through the suit."
There's a clear purpose here. Humanoid robot testing of garments intended for hazardous duty. I doubt anyone at Boston Dynamics or the DoD are sitting around pondering how they're going to cloth the obscene masses of robot humanoids they plan to produce.
[1] http://www.youtube.com/watch?v=mclbVTIYG8E
I'm sure they wanted their proof of concept out just to be the first robot with clothes, but just didn't want to risk damage.
Source: been there, done that, don't tell OSHA. :P
http://hackaday.com/2013/03/29/bigdog-throwing-cinder-blocks...
Certainly, the power divide between the government and people (inside or outside of its borders) continues to grow. The scary thing to me (in a distopian future) is that robots won't hesitate to do as they are programed. Soldiers sent in to "subdue" their own population might at least think twice.
The advantage of dynamic stability is of course that your robot is constantly in danger of falling down, that is its natural state, so trying to knock it over doesn't really change what its doing a whole lot.
I did not think I would see dynamic stability in a humanoid shape though (power off condition is laying horizontal :-). Watching it is pretty unnerving. I'm impressed they have it working.