Slashdot Mirror

Yep...

http://www.rethinkrobotics.com...

Even if it costs $250,000 to replace each worker, consider that you're getting a 24/7 machine, so really it replaces 3 workers (for a 24/7 fast food joint).

It never calls in sick, never gets lazy or stupid, it won't spit in the burgers, and it only has to be trained once.

Oh by the way, install 4 of them in each location, times 10,000 locations, and now you have 40,000 of them, and once you train one, all 40,000 of them know how to do it.

Your training costs just went to nearly nothing.

http://www.rethinkrobotics.com...

Not really...

Starting at $25,000, Baxter is quite reasonable. He isn't nearly as fast as a human, but consider that he'll last for years, works 24/7, never shows up late, never calls in sick, and uses pennies of electricity while his meat based competition is now going to cost $15/hr.

So a $25,000 robot replacing 3x workers in a 24/7 McDonalds is $90K, except that employees cost more than you pay them, it is really closer to $120,000.

You can buy 5 Baxtors for that price.

A quote I like:

A tenth of the speed is a bargain when it is one hundredth the price...

Use Baxtor for 5 years and it is FAR cheaper than hiring 3 employees for 5 years. Also consider that if you're McDonalds, you likely will pay less than list price for him because you can use 100,000 of them.

You also have almost no training costs. Show Baxtor how to do it once, and all 100,000 of him now know how to do it. Perfectly, the same way, every single time.

Replying to my own post...

http://www.rethinkrobotics.com...

Found it...

It costs $25,000 (down from the $30K number I remembered), it doesn't require special programming, you just show it what to do.

Scroll down that page, look at what it can do, including putting items in boxes.

Now it might not be ready for what Amazon needs, but given the size of Amazon's workforce, their budget, and the possible savings from having a million such robots and not needing to "surge hire" for the holidays, clearly this is a goal that Amazon will continue to peruse.

---

A tenth of the speed of a human is not a problem when the hourly cost is 1/100th the price of a human.

https://www.youtube.com/watch?...

Found that video as well, it is called "Humans Need Not Apply" and it covers some of the ideas for the future of such robots.

This guy here can make a sandwich and many other things that humans can do for about $4/hour.

The 2012 interview was more informative:

" Indeed. We don't mean "common sense" from a Marvin Minsky-like strong AI perspective. Baxter's "execution" application consists of a series of behavior-based systems. During "training," the robot detects task-relevant features and uses it to build up the behavior based system.
For example, let's say a user is training the robot for a pick and place task. During the "pick" phase, a user places the gripper above an object and closes the gripper. The force on the gripper is detected by the robot. Our "training" application detects this sequence as "the robot is grasping an object"... so during "execution", Baxter won't proceed unless it actually detects an object in the robots gripper. Thus, if the object fell out, it would stop (or do something else). This is different from how existing industrial robots work -- they'd just merrily continue the pick-and-place without the object.
Collectively, these "behavior primitives" are assigned and composed, ie. "learned", during "training" by having non-technical users directly manipulate the robot rather than programming it (which is also possible for those inclined). This gives the robot an air of common sense."

This is useful, but not that intelligent. Take a look at these PR videos to see what it can do. Basically, it can pack and unpack things, and move them from one place to another. It's not good enough to assemble much of anything. Plugging in connectors to assemble a phone? Not with this machine and software.

http://www.rethinkrobotics.com/products/baxter/

I've been tempted to get one as an investment ($22K) to learn to write apps for it, but I don't have the time right now doing other work. But stuff like Baxter is clearly the future...

And, while we "rethink robotics", we need to "rethink economics" (including a basic income, an expanded gift economy, improved local subsistence, and better democratic planning), like I talk about on my site.

Hi, I'm a professional roboticist. Here's some super awesome (often deliciously job-crushing) automation systems off the top of my head:

Automated fast food preparation. Yes, that's right. Those jobs are going.
Safe, easily reprogrammable robotic factory line workers. No light curtains. More cost effective than a minimum wage US worker and still improving.
Automated chemical solution preparation. This normally eats up the time of lab researchers using their PhD to essentialy do high school chemistry that's standard grunt work.
Modular biological lab automation systems. Similar to the previous one, this eliminates a bunch of grunt work that lab researchers normally have to do themselves.
More laboratory automation.
A fairly high-end pick and place machine assembling PCBs. Shenzen eat your heart out.

Industrial robots have been expensive semi-custom products for decades, and there's no good reason for that.

This product isn't going to replace the expensive semi-custom robot systems; that is not their target market. This is enabling automation in lower-speed, lower-volume, low-complexity tasks. Look at the specs listed at the bottom of this page on the Rethink Robotics website.
- 8-12 pick & place operations/minute (total incl. both arms)
- 5 lb. payload per arm
- 1 m/sec arm speed

So they won't be competing with the following "expensive semi-custom products":
- high-speed pick and place (i.e. PCB surface mount components) - cycle rate 10 to 20 times higher than that
- anything high payload
- anything with a complex custom end-effector (counts against your payload)
- I can go on.

Ultimately you get what you pay for; there will be tasks where these are suitable, but they will not be replacing high-power, high-speed, custom-engineered robot/automation systems any time soon, as they aren't intended for/capable of those tasks. I'm sure there's a market niche for these, but is it going to transform the world of industrial robotics? No.

That seems unlikely; robots are expensive, but much more importantly, robots capable of doing many unskilled tasks simply don't exist yet. Fully automating humans out of industry sounds much easier than designing a consumer-oriented mass market robot because you can at least have a semi-controlled environment. The most user-friendly robot I have seen is Rethink Robotic's Baxter (and that's their page on it) which is ~$22k and can be trained to perform repetitive motions. The really fancy part is that the repetitive motions can be at the level of "pick up object from here and place it there" instead of "move arm, grasp [hopefully where there's an object but really who knows], move arm, let go".

Humans aren't going to be entirely replaced any time soon, but unskilled labor will likely be eliminated as category of job before anyone thinks about designing a consumer robot much fancier than a Roomba. Before such a product is invented, very few people will have jobs, so not many people will be buying them.

Baxter