Latest insanity from Boston Dynamics


#61

I think I’ve got it! What if we put some kind of device on the door, accessible only from one side, that stopped the door from opening while it was engaged? I don’t know what you’d call such a thing, but I’m sending in the patent application now!


#62

How about self-driving delivery van pulls up, out pops the robot dog carrying your delivery, it trots up to the kitchen door, uses an unlock code, opens the door, and places the delivery in the designated area?


#63

Where “delivery” is “needle full of polonium” and “designated area” is “your eyeball”.


#64

[insert Millennial meme about wishing for the death with the ease of Prime delivery here]


#65

I’m not suggesting that, I’m directly stating it. What evidence do you have that people want or need these kind of robots? It’s entertainment, there is no demand whatsoever.

Real, working robots look nothing like this. They look like forklifts without a seat or mechanical arms or remote controlled vehicles. Walking, animal looking robots have no demand or use in our society so far and I can’t imagine where that demand comes from.

Look at security. There are no security guard robots like in mid 20th century fiction. What ended up happening was video cameras, remote storage and facial recognition.


#66

They got bought last year by Softbank in Japan, who does make robot products. This is after their contracted work for Darpa and being bought and eventually sold by Google. I’m assuming something in the portfolio drew the eyes of the Softbank folks, so SOMEONE thinks there is something marketable there.


#67

Google sold BD because they had no marketable or usable product in the foreseeable future. Basically the head of Android got all excited by them and brokered that deal, then when he left Google execs went “what have we done?” and sold that turd.

Don’t get me wrong, I love their robots, I love BD, I love their videos. But they should be in movies as props. They are not going to be releasing a widely available robot like the ones in these videos… probably ever. Robots don’t look like this, they are custom designed for their role. This is entertainment.


#68

From my understanding the DARPA contracts never went anywhere. They were originally trying to use them as pack carriers out in the field but they required too much power and their engines were too loud to be practical.


#69

There’s no marketable product value at present, for sure. That robot in the latest video is probably being remote controlled for choreography purposes, and in any event cost a zillion times as much as installing a door-opener or even just hiring some human to stand by the door and open it manually. It has no industrial use, because factory floors are not whimsically set up with doors to block the conveyor belts, and because other applications like delivery can’t possibly make use of such an expensive item as a BD robot anyway.

But the whole point of this is it’s the technology that has value, not the prototype. The tech is what lets it walk without falling over when it nudges something, and what lets it automatically lock onto the door handle with an off-the-shelf robot arm, nudge it open, and hold the door for the other robot. This kind of thing will be enormously valuable in the near future – not the usual 3-5 year time frame, but 10-20. Having the R&D expertise and the patents will be of great value when the products mature. Just not worth buying or selling product right now.


#70

How about for people who have lost limbs?


#71

The people who could afford it can also afford an army of servants for the same price who could do a far better job. Once again this is a demo prototype in a staged situation. It’s a hugely impressive tech demonstrator, but it’s not a salable product for that kind of application at present. It’s a clear indication of what the future will bring, however; eventually products will be available at a reasonable price. The question is when this will happen. I estimate 10-20 years, but that’s just intuition.


#72


#73

Lol. R2 D2 fell down. Or perhaps a Dalek.


#74

No. Can’t you see it’s obviously swimming.


#75

It is merely cooling itself Battletech style before it unleashes its machine fury on the weak meat bags it has lured close.


#76

I was going to link that very robot, more about how it is primarily used to run off homeless people 24x7 and the furor that caused, but the shot of one taking a bath is so much better.


#77

The thing that impresses me the most about the robots movements are how it uses its tools in an unusual way for a robot, something closer to what a human would do. This might have something to do with how BD does their programming, as I’ve read that they work much more with manually shaping movements as opposed to some sort of neural net learning approach. The parts that stand out for me are how the robot uses a leg at 27s in to hold the door open (like a human would a foot) while it changes it’s grip from the outside door handle to using the side of the gripper (which seems unusual in itself) on the interior side of the door, then releases the wedging leg as the arm again takes on the force of the door.


#78

I don’t know how much of that demo is direct human remote control, how much is hard-scripted action for the demo, and how much is “natural” generic behavior that would enable an autonomous robot to adapt on its own to open any door with a similar handle. I suspect it’s not really all that smart, but I have no information offhand either way.


#79

I dunno, vision rec has come a long way and the BD guys are aces. I expect it’s not a staged demo, but maybe.


#80

BD claims it’s entirely dynamic autonomous behavior, see this older video, where they try to slam the door on the robot: