Foellmer demonstrated what happens when you stick your arm into the path of Baxter's pick-and-place operation -- it stops right away. It also won't pinch you if you get your arm or leg stuck inside its area of operation, such as under one of its moving arms. At first I let Foellmer risk his limbs, not mine. Then I let Baxter's arms run into my arm. The robot's force-detecting motors stopped it immediately.
Being also a materials person, I immediately noticed that Baxter has mostly plastic surfaces, not hard, metallic ones. That's highly uncommon in industrial machines, although not unusual in some service robots. It's also another thing that makes this robot safer.
The second major goal was to make Baxter simple enough so human workers can train it to do easy but repetitive tasks. In other words, it can be programmed by people on the floor, not those with advanced engineering degrees wearing a special software pendant. The fact that it doesn't need complex programming also means it's simple to integrate into existing automated operations, although many smaller manufacturers don't have existing automated assembly lines anyway. Most of their assembly cells are inhabited by humans doing things by hand. But the simplicity of programming makes changing Baxter's tasks much easier than is usually the case in industrial robots. We've discussed that excruciating complexity on several Design News comment boards.
Rethink Robotics' Baxter. (Source: Design News)
Baxter does not move large parts: It's got a payload limit of five pounds per arm. It's also got a small footprint, weighs about the same as an average-sized adult man, and can be moved around fairly easily. It's clearly not meant to do everything, and doesn't do the kind of high-speed pick-and-place on display in Anaheim by major industrial robot makers like ABB. The other revolutionary thing about Baxter is its open-source Unix-based OS, ROS (robot operating system). That plus an SDK to be released later this year will help open up the robot as a platform for development.
Baxter's $22,000 price tag is much lower than the typical two-armed industrial robot, which should make it more appealing to small companies that want to start using automation. It's already been beta tested at a couple of customer sites, Foellmer said, and has just started shipping.
I know exactly what you mean, Chuck--actually, it looks more like what's called a fantasy weapon, which are more extreme versions of actual (usually medieval) weaponry used in both historical and fantasy movies and some role-playing/re-enactment games, and are represented in some video games.
Tim, Baxter isn't really designed to handle fine pick and place movements such as is needed in small-parts electronics assembly. Those are very sophisticated, expensive, precise machines. It's targeted at less precise movements. It's also designed to work alongside humans more than to interact with other robots.
Good article, Ann. Looks like the trade show had a lot of interesting products to keep you busy.
You've got two great concepts here, but keeping them separate might be a good idea. Imagine Baxter with that bone rasp in each "hand" and an angry face on the computer screen!!!
Seriously, as for the concern about differentiating between a person or a part, I wonder if the flesh-sensing technology used in saws (i.e. table saws) would be able to be integrated into the "skin" of a robot to help it identify humans. Since the saw companies are resisting using the technology, perhaps the robot industry would be able to incorporate it.
But the real point is that 3D printers can make complex shapes that would be too costly (translated: impossible) by other methods. I can imagine that bone cells would really gather 'round this object and build new bone. Additive technology will help us build shapes previously unattainable.
@CLMcDade , That "skin detector" used in the sawstop system would not help in a robot system because it uses a resistance principle, not a touch principle. And the reason that the saw companies are not rushing to adopt this system is that it has a few very big shortcomings, including a very expensive reset process and a propensity toward false triggers from wet wood and nails.
The two steps to make a robot safe for humans to be around is to slow it down to human speeds, and to eliminate pinch-points. By no means a trivial task, but certainly an achieveable target.
Interesting use of safety technology. From their website, Baxter contains sensors and software protocols that detect people within contact distance and trigger the robot to slow to safe operation speeds. May be that the robot sets up programmable safety zones on sensor inputs. Every motor can also be "back driven" in order to comply when unexpectedly pushed backwards.
Thanks Clinton. I did not think of Baxter wielding the bone rasp--I take no responsibility for others' imaginations! OTOH, Chuck, pointed out that it looks something like a medieval weapon, so I can understand the association. That's an interesting idea about flesh-sensors; I didn't know about that. Sounds like a good cross-app possibility. Hope Rethink is reading these comments...
eafpres, it's pretty simple. if something larger than a part--like the human body--gets inside its working zone, it stops. This is determined by its sensors. Also, if you bump into it faster than it can respond, it won't hurt you because of its softer surface (plastic) and its considerably lower force, compared to other industrial 'bots. More details are available on the website.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.