Rethink has been one of the pioneers of “collaborative robotics” — unlike traditional assembly-line robots, which are programmed to do one task and usually walled off from human workers for safety’s sake, bots like Baxter are designed to work safely alongside humans, and can be moved around a factory floor and shown how to do new tasks by workers. That makes them better suited for smaller factories and workshops. And they’re relatively cheap.
Baxter’s diverse skill set has already made him the star of many a think piece on how robots will impact the economy. But what’s fascinating about Crooks’ work is that she and her team are trying to get the bot off the factory floor and into your living room — for her thesis project, she’s attempting to turn him into a kind of robotic home health aid, something that could fetch things for elderly or disabled people and allow medical worker to check in on them remotely. Considering that health care aids are often held up as an example of the one of the professions least likely to be impacted by automation, I was curious about what kind of potential Crooks saw for their application. We spoke about the current state of her work and the possibilities — and limitations — posed by bots like Baxter.
MPRA: So how did you become interested in robotics?
WC: When I was eleven-ish, my step-dad sent me a video of someone demonstrating a prosthetic arm. No one in my family was particularly into science at a high level — my mom taught math and science to middle school kids, but no one was into engineering or anything like that. But I thought it was really cool, the more I learned about it, the more it seemed like playing for a living and like something I’d want to do.
MPRA: Tell me a little bit more about your experiments with Baxter.
WC: Right now we’re developing with a setup with a team from Rhode Island where we’re thinking about mounting him on the carriage of a motorized wheelchair, and Baxter can drive himself or you can control him remotely. We’re really focusing on making Baxter tele-operable, and figuring out what tools we can design for Baxter. As capable as Baxter is, there’s some fairly significant limitations when you’re thinking about putting him in someone’s home. Each arm can only pick up five pounds — if you’re putting him in a situation where someone might be falling, or might need Baxter to help move something, we really need to overcome that limitation. So it’s an interesting mechanical engineering and robotics problem.
MPRA: So what use cases are you testing right now?
WC: Right now we’re focused on one-arm manipulation — we’re having him pick up coffee mugs. No coffee in them yet. Maybe soon we’ll add something weighted that’s dry. And probably is plastic, so they won’t break if he drops them. [laughs] But we’re teaching him to manipulate lots of objects: Eyeglasses, tissues, pens and pencils, books — those are actually kind of hard because of their shape. For us it’s pretty easy to get [our fingers] under things to grasp an object, but a robot can’t always get under the object. Working on this has really made me realise how amazing our hands are.
We’re also thinking of adding interchangeable hands [to Baxter]. There might be some applications that you really want an actuated, finger-grip hand for. So maybe Baxter could have a row of potential appendages attached, and depending on the task he can select the one that’s most appropriate. We’ve been calling it his Bat belt. Kind of like we would have a tool box.
MPRA: I suppose that’s one of the advantages of robots — we can’t just chop our hands off and switch to a hook when even when that might be more advantageous.
WC: Right. It’s important to use the bio-inspired stuff, but also there are things robots can do that we can’t do. Some bots can turn their hands 360 degrees, which we can’t do. Even though sometime it’d be really nice not to have to sit there with a screwdriver going like this. [Mimes making a quarter turn].
MPRA: Do you find for you as a designer that that’s still a mental obstacle you have to overcome? If I were designing a robot to screw things in, it might not occur to me that the bot can spin its hand all the way around.
WC: I think I’m getting better at doing that. I remember coming out of undergrad, my senior project was a robotic grasshopper, which we designed out of metal and plastic. And now I’m aware of so many more ways to approach that, using soft materials. I think being part of the soft robotics program [at Tufts] really expanded my view.
MPRA: It was thinking about this watching the recent DARPA contest — the idea with that was you wanted emergency bots which could operate in areas designed for humans, so they should be humanoid bots. But looking at the video of some of the ways they failed, it made me wonder whether a non-humanoid bot might have been better to tackle some of those tasks.
WC: I gave a recent talk where I presented with someone from Rethink [Robotics] who talked about Baxter, and I was more on the research side. Afterward people were asking questions about what Baxter could do [around the home] and they were really fixated on stuff like having him shovel the walk for them. I was thinking, well maybe it’d just be better to automate your snowblower. A humanoid robot would have the same problems that you and I have — slipping and falling all over the place, potentially breaking something.
MPRA: Many economists seem to think that fields like home health care will be one of the last to be conquered by robots.Yet having him help out people in their homes was one of the first things you thought of once you had access to one. Do you think people don’t appreciate robots’ full potential?
WC: I think [our research team’s approach is] different in that we really want to keep people in the loop. There are things that machines aren’t good at yet, like vision, identifying objects. [And so I think for now you’ll need to have a human controlling the bot to help with things like that.] But there are so many people out there who are elderly, who are mobility impaired, who are losing out on social interaction, which is really a key part of who were are as people. And I think these bots present so many opportunities to improve people’s quality of life and keep them independent in their homes….there are a lot of ways for robots to improve our lives, if we can improve the interaction between humans and robots.
MPRA: It feels like in some ways the field is plunging ahead with putting all these bots in factories and homes at the same time as we’re just starting to do the research to understand how humans interact with robots.
WC: I think there are lot of people who don’t really understand the state of robotics — I keep getting asked, when will they become self-aware? And we’re so far away from that. Robots are for the most part, really dumb. Even the smart ones.