Google Has Made a Mess of Robotics
bloomberg.com
bloomberg.com
Because of the nature of economics (vendors are either selling parts or complete systems), no one had incentive to work on a reusable OS / framework. This meant every lab project and product would start from scratch and rewrite the same stuff. From seeing this at a research lab ~2011, it absolutely impeded progress.
Grad students shouldn't have had to repeatedly reimplement all the utility bits just to test their ideas on a far more specific robotics problem.
It's a good talk, and probably the first time I've seen Boston Dynamics robot portrayed in a positive light. The videos they have released in the past, has always given of this Terminator vibe.
Why? Well, when I leave work, I spend 45 minutes in the car, and another 45 minutes preparing dinner. With self driving cars, that will not change. But with self cooking chefs, that would add 45 minutes of quality time to my day.
The difference is that a self driving taxi will be far cheaper than a taxi, partly because it can be shared.
A robot that prepared food already exists, it's called a factory, but food is far more variable in terms of people's desires.
The economic impact of car accidents is almost as staggering as the numbers of injuries and lives lost. The total economic impact of car crashes is a whopping $871 billion in the U.S. (http://www.thedrive.com/sheetmetal/13792/car-accidents-cause...)
The revenue of the entire restaurant industry is less than that total economic impact ($799 billion http://www.restaurant.org/News-Research/Research/Facts-at-a-...)
The Economic and Societal Impact of Motor Vehicle Crashes https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/...
Note that these numbers are made up to a certain degree, much more so than the restaurant industry numbers, and so they are not necessarily comparable. The NHTSA puts the value of one life at $8.86 million (in 2010 dollars) [p. 114 of the report]. For 40,000 deaths, that's already $354 billion. They then add other costs, too, which are also hard or impossible to measure.
I know this is a normal thing that normal people do; I know people who do it, but it is something I personally find incomprehensibly difficult.
Truck driver is, IIRC, the largest single job description in the US. That's a huge target market for self-driving technology and a massive societal problem that's looking when so many people are left jobless.
It makes more sense for the dinner cooking robot to be making thousands of meals a day than just two for you each day.
We are building a network of automated food preparation and retail service locations for high density mainland Chinese cities, with a view toward international expansion.
Your statement concurs with the results of a study published in PNAS and reported in NYT recently regarding cross-demographic improvements in happiness resulting from increased spending on time-saving - http://www.pnas.org/content/early/2017/07/18/1706541114.full and https://www.nytimes.com/2017/07/27/science/study-happy-save-...
Even in the cost of buying the vehicle, it is already highly automated, how much can more robots decrease the price of the car?
Robotics in general may not have strong barriers.
Transformers are cars that drive themselves.
I think it's a case that will be reinforced with the first viable self driven car.
Even if Google did exactly nothing with robotics (which seems unlikely), the infusion of capital to the people with expertise is good for the field.
As a result I would expect to see a flourishing of robotics research in the near future, I'm just disappointed that they were stuck there for 4 years in the interim.
It makes me wonder if companies that are attempting to acquire smaller companies might not just be better off leaving them in situ and acting in more of an investment role than an ownership role. Buy 52% of the company for $x/1.9 instead of the whole thing, and in doing so provide them with funds but not interference.
Further, you are also asking a corporation to admit that it's not as good at deploying capital as the smaller company. That's may be true but is a tough position to take.
I'll take the capital infusion with 4 year delay versus no capital.
That's literally what they're saying by acquiring though. They're saying "We could use this money internally to build a clone but your company is operating at a discount to what we can do ourselves"
And yes, certainly, capital is good either way :)
The choice is something like $5M acquisition or $1M building it + 2 years
For a large corporation, it makes sense that money is cheaper than time
I can see where your idea has merit in the long term, google seems to have set back robotics by 4 years or so in the short term.
There's a lot of money chasing some ideas. I wonder if tech would develop faster if there were a little less money involved?
Robotics is probably on the edge of this, between software (where it's probably true) and less-smart hardware (where it's probably false).
Having worked at a robotics lab, budget absolutely comes into play when you know what you need, but it's a custom part from a single vendor.
Not a flattering narrative for Google's public image.
I don't fucking understand what the fuck is going on? every group who complains about some of form "ism" ends up doing it.
That's the ageist statement. Google doesn't hire three year olds.
They proceeded to tell their engineers to "cut it out" and started creating parameters like, "no more bipedal robots" and "stop scaring the public." And, actually, their engineers ignored them and kept doing it, which annoyed Alphabet further so they put additional pressure to stop work in that direction. Though, the irony is that this led to a pretty awesome terminator bot that uses wheels, moves ridiculously fast, can jump and balance better than their bipedal robots and therefore become your friend faster.
I don't personally feel they've held back robotics as an actual field, because the progress continued internally in spite of Alphabet's mismanagement and fear. However, Alphabet suppressed those robots from becoming mainstream products because they weren't the products they wanted and they were very sensitive about the impact on their public image.
That division fully had the capability to release robots that could help people or industries in various ways; military (pack bots, soldiers), police (peacekeepers), steal your job bots, medical (bots that help lift people out of bed, or keep old people from wandering out of their prisons), pets (pet dogs, cheetahs that can hunt down your ex with speed and accuracy, a spider that can climb in bed with you and inject you with life saving chemicals that extends your life before giving you a hug goodnight), companions (definitely think sexual here, perhaps even combined with some of those pet ideas), and so on.
It might have begat the dystopian future that keeps many of us, or just Elon Musk, awake at night. And, if that's the case be thankful Alphabet couldn't manage to get much into production.
The only real way they've held back robotics is by never releasing anything, beyond previous commitments of course.