Rod Brooks Reveals Baxter, Programmable Industrial Robot for the Masses | Hacker News Reader

Rod Brooks Reveals Baxter, Programmable Industrial Robot for the Masses

xconomy.com

72 pointsrobertbud125 comments

As a professional roboticist, this is really exciting. It could have an even bigger impact that the Kinect (and other depth sensors). For the last few decades, actuators have been the dominant cost of building such a robot. Comparable arms routinely cost ~$100k (from the PR2, Meka, ABB, Barrett, Kuka, etc). Essentially, Baxter represents a 10x cost savings, taking the per-arm cost from ~$100k to ~$10k each.

Actually... Rod is my "academic grandfather." Yesterday I had the opportunity to interview him for Hizook's feature article:

http://www.hizook.com/blog/2012/09/18/baxter-robot-rethink-r...

So much for yesterday's HN frontpage article about hardware being dead...

I've been curious lately about how robotic arms are designed and built. Do you have any beginner resources you could point me to?

Don't want to nitpick, but the article says "Baxter will sell for $22,000, not including its base and hands, but including a software subscription and warranty." So not really a 10x cost saving.

Wow that's really cool. How far away are we from having one of these do dishes?

Optimized high-performance robots for washing dishes, adapted to the consumer environment, have been available for almost a century. They're called, quite aptly, dishwashers.

Ok, when will it load and unload the dishes?

Right, and for a bonus "hand" wash things that don't fit in the dishwasher.

Get two dishwashers and put your clean dishes in one of them.

Put a sticker on it marked "clean". Need a plate? Take it out of the "clean" dishwasher. Got a dirty plate? Put it into the other dishwasher. Other dishwasher full? Turn it on and move the "clean" sticker to that one.

Double buffering with dishwashers turns them into magic cupboards. I don't know why most kitchens only come with one.

It sounds good but I wonder how it would work in practice? What if you have a lot of guests over and use more dishes than usual?

What about items that you don't use fast enough? They'd stay left in the clean machine and you'd have to manually put them away before it turns into a "dirty" machine.

Maybe you really need more than 2 machines?

Because if you have a family of four, all of your dishes wont fit in the dishwasher.

Double buffering with dishwashers turns them into magic cupboards. I don't know why most kitchens only come with one.

One is enough if it's a Fisher & Paykel DishDrawer. Two independent dishwashers in one: http://en.wikipedia.org/wiki/Drawer_dishwasher

For storing things it probably wouldn't be big enough for more than a couple of people though. Also, magnet is better than sticker.

Thank you! This is a nice hack. Going to implement this at home.

> Put a sticker on it marked "clean".

The sticker/magnet doesn't work.

What works is maintaining the invariant "if the door is locked, the dishes inside are clean". If you open a locked door, you either empty the dishwasher or relock the door.

It would be great for consumers if these came with an app store where you could buy programs to do different tasks around the house (wash dishes, cook a specific meal, etc). If it was useful enough, it's not crazy to think that a family would spend 20k on one (same cost as a car).

And a lot cheaper than a maid.

Yes, or folding and putting away laundry. The bane of modern existence! I'll take a hack for that if you've got one.

Single page: http://www.xconomy.com/boston/2012/09/18/rod-brooks-and-reth...

The $22K price is for an incomplete system. The two missing pieces (hands and base) sound pretty integral to the system as well. What is the point of listing a price of a useless configuration?

The arm itself is just the platform you build on to implement the end device. The hand and base are specific to the application.

End effectors depends on the task, many are already available on the market. Some applications might not even require an active end effector at all. Same thing with the platform. Research labs who want to explore user interaction with safe, compliant arms without designing their own now have a much cheaper option that before.

Unrelated to the content but if you do insist on chunking your articles up into pages, at least break them at paragraph boundaries! Or even sentence boundaries - breaking them mid-sentence is atrocious.

The reverse sensor programming reminds me of cg animation auto-keyframing.

Yes, it is presumably doing some inverse kinematics in there somewhere. I wonder how they stabilised it.

Does anyone know how the arms are designed? Do they just have big servos?

Missed opportunity: name it Dexter.

Reply on news.ycombinator.com