Seems ironic that robots in science fiction were envisioned as menial servants able to do physical tasks too hard or unpleasant for humans, whereas this robot, with it's sleek 35lb body and advanced communication ability yet with no arms seems better suited to play the role of master in a world in which many cultures have regarded abstinence from physical work as a mark of superiority (for example, in asian cultures, very long fingernails was a mark of class because they could only be maintained by a person who didn't have to do manual work.)
It's for mobile telepresence. E.g. you send this to the factory in China instead of flying there yourself. It has cameras and audio and a fancy laser pointer they built themselves. You drive it via the Internet from an ordinary computer.
That would be a bit creepy though. Imagine a factory of chinese workers, all sewing and this this silent rolling monster drifts slowly by, and they all stiffen in fear and work extra hard...
iRobot made something similar called the "coworker". It was a spectacular failure. It was shorter and ( i think ) a bit smarter about remote obstacle avoidance. They asked around $5K.
They never answered the question: why is this better than a conference call? This is doubly important considering it has no arms.
I'm certain there is a market for this or a similar kind of robot. It needs to have compelling remote presence, and it needs to be cheaper than $5K.
Adding up motors, computers, cameras, the chassis, plastic modling, etc. can get expensive fast, especially if you haven't ramped up production somewhere fast & cheap like China.
Why better than a conference call? It has to be it's mobility. (Although, didn't they mention it needed to be tethered? It has to be because it lacks batteries.)
I think that in situations where data needs to be sucked up in a mobile way but not in a way that the environment needs to be manipulated (no arms), it'll work. Well, when it gets batteries, it would.
"Unlike monty, QA is 100% battery powered, and gets 4-6 hours of runtime on every charge. Monty's pneumatics required it to be tethered at all times. Also, its much lighter at around 35lbs. Monty is too heavy to pick up."
This robot makes small jerky movements. I think the others are much more natural in appearance. Perhaps it is more of a hardware issue than software, as it does seem to have pretty good balance.
Truer than you know. The reason the Asimo's gait looks so smooth is that it's preprogrammed. It doesn't balance dynamically. It looks smooth for the same reason animation does.
That's not technically true. Asimo has preprogrammed paths, but not motor movements. Even in a semi controlled environment, a preprogrammed biped would fall.
The difference is between electric motors and pneumatics: today's pneumatics are just jerkier.
And static balancing doesn't mean that it is preprogrammed, just that at any point (unless it is running), all movement could stop and the robot would stay standing.
If humans stop moving, we collapse.
I guess the best reason to say Asimo is not more natural is that no animal in nature is statically stable. All animals with legs have actuators with the characteristics of pneumatics and springs (in humans, tendons are springs).
We can also turn to efficiency. The electric motors on Asimo will _never_ be practical. It takes too much energy for the bot to ever be sold as a product. No animal in nature has a 15 minute battery life :)
static balancing doesn't mean that it is preprogrammed
Really? That's what I was referring to when I said the gait was preprogrammed. I assumed that static balancing meant that the robot was only willing to traverse certain paths through the n-dimensional space of possible limb configurations, and that since these could be calculated in advance, they would be. But since I know you know about robotics, I'm willing to be convinced otherwise.
There are constant adjustments made by the dozens of motors in the legs and torso to keep balance, based upon sensors like accelerometers and gyroscopes. That's not preprogrammed.
My point about a reasonably controlled environment is that in the real world, even on a "flat" stage for a show, there are enough irregularities to make complete preprogramming hopeless.
I don't know enough control theory, but it could be stated that an open loop controller will be doomed to fail for a biped.
Higher level, there are sequences of commands that yield a step forward or to the side. That is almost certainly preprogrammed. When you own an asimo, the api they expose is actually really limited. For example, to lift the arm from the side to the air will only have a few states in between, and you choose which one you want. This is not low level control at all. They do this because they know Asimo is a marketing machine, not a real robot, and they don't want grad students making it fall over on youtube.
On second thought, a flamingo standing on one leg might actually be statically stable, and there might be other examples. But flamingos are weird, amirite?
I just saw it this morning... I have to say, it honestly was one of two things that gave me a "holy shit!" moment at CES. First of all, they were the only exhibit really getting much attention in the Sands; and they were getting a TON of it.
Secondly, the description doesn't do it justice. They were doing a demo of it, and there was just something about the way this thing is put together that makes it feel like it came straight out of Star Wars.
Huge props to Trevor and crew for making something so cool - I hope you guys found some deals at CES!
