Terminator-style robot busts leg in martial arts demo mishap Hong Kong University robotics researchers ended up with egg on their faces today when their new $2m walking robot tripped over and broke its leg while trying to strut its funky stuff at its first public performance. HKU’s prize robot – called Atlas and created by Massachusetts Institute of Technology spin-off Boston Dynamics …
Tai chi is very popular among old people here, you can see them in the parks, early in the morning, practising. For those not familiar with this martial art, it is characterised by slow, smooth movements. It is martial in origin, but is a form of gentle stretching exercise now.
Don't be fooled by the apparent slowness. First of all, it's a lot harder to do movements slowly than it is to do them quickly - so slow means you truly master the muscular difficulty.
Second, I once saw a Tai-Chi master demonstrate a movement in slow and quick modes. I guarantee that the quick mode was quite impressive, and I do not doubt that an opponent would have been seriously injured by that flurry of activity.
According to Boston Dynamics, Atlas is a high mobility robot designed to negotiate difficult outdoor terrain.
Ah, there's your problem, right there...
It should have been designed to negotiate tricky indoor terrain, you know like highly polished lab floors, loose rugs, spilt cups of coffee, that sot of thing.
The other hot news is, "Transformers director attacked by 'zombie' triad wielding aircon unit"
http://www.scmp.com/news/hong-kong/article/1333800/transformers-director-attacked-zombie-triad-wielding-aircon-unit
Beware, the Chinese have more weird martial arts than you can shake a stick random improvised weapon at.
quote: "when you look at this machine and the heavy cables powering it, just remember. A fit human can run a marathon consuming a pound fat in energy...."
This got me highly interested to see just how different the energy requirements are between organic runner and mechanical walker, so I thought I'd do a little digging and some maths:
According to Wikipedia fat has an energy density of 37kJ/g, so a pound (~454g) would contain approximately 16.8MJ available for metabolisation.
Used in an hour, that would be an ongoing power requirement of 4.6kW (16.8MJ / 3600s), although a quick search suggests an "average" time for the London marathon is more like 4 hours, so I'm going to take 1.2kW ongoing power usage (1.16 rounded up) for your human running a marathon. 1.2kW delivered via 240VAC mains electricity is around 4.8A current, which is actually a significant amount.
In fact subsequent searching to see if anyone has actually measured the efficiency of humans, brings up that human (skeletal) muscles are about 18% to 26% energy efficient apparently. If the actuators on the robot can hit 30% or more, it may actually be more energy efficient that we are ^^;
Keep in mind that the energy content in fat is measured as chemical energy released in the form of thermal energy based on combustion. So humans convert that chemical energy to motion at 18-26% efficiency. The robot is using electricity to generate motion, which is already a very high quality form of energy. For 1 J of power from electricity, you'll typically need to combust 2.5 J of chemical energy in some fuel to drive a heat engine (Brayton or Rankine cycle), so to properly compare the efficiency, you need to start from a chemical energy source in both cases and not give the robot the advantage of a electrical-to-motion efficiency while discussing human's chemical-to-motion efficiency.
Hmmm... 1/2.5 is around 40% efficiency (almost identical to the losses quoted for humans converting food energy to ATP, bizarrely enough). According to a quick search, some unsubstantiated figures would put electrical motors at an alleged achievable efficiency of 70% to 90%.
Putting the 2 together, 70% of the 40% would make an (apparently lower efficiency) electrical motor approximately 28% efficient, based on hydrocarbon burning in a heat engine. If you wanted to take the upper bounds for that quoted motor efficiency, it would be around 36% overall.
Taking extra losses into account (friction, startup losses etc.), I still reckon the 18-26% efficiency quoted for humans sounds achievable (and possibly even exceedable) by a humaniform robot. It would also have a similar CO2 output if you are truly running them both off the same amount of organic compounds ^^;
I can't say that this was the conclusion I was expecting, to be honest...
This post has been deleted by its author
Isaac Asimov suggested that human form robots were ultimately the most practical and versatile as they could use human tools, drive human vehicles and navigate human environments. Although ultimately more complicated to design, a human form robot could be used for almost any task currently performed by a human without redesigning our whole world around it (at incalculable expense).
Big Dog serves a vary narrow, niche purpose - it can't pick up and fire a weapon, drive a car, dial a phone or even turn on a light switch. In fact, last I saw, it couldn't even perform all the tasks of a dog - it cannot fetch, it's more of a Donkey than a Dog.
This video is very scary, especially Atlas manically crawling through a corridor.
Reminds me of the homicial robot from 1980's Saturn 3.
The sexbot industry has a long way to go, if this is an example of current tech. It has the "lying there" part down pat, but the "looking pretty" part? Ehhh...NO. Not if were the last bot on Earth. Clearly, DARPA has the wrong priorities when it comes to "supporting our troops". I believe our entire military would applaud ATTRACTIVE, disease-free sexbots to "support our troops". This bot - ewwwww !
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
