Japan’s Laundry Folding Robot Is Taking Over Your Closet – Seven Dreamers

It’s often surprising to discover which problems are hard for AI. We hear stories about artificial intelligence being better than the most skilled humans at go, chess, Jeopardy, and better than many at driving a car, and we assume that computers will be as smart as we are very soon.

Then we discover how hard it is for AI to fold the laundry.

Shin Sakane and his team at Seven Dreamers have been working on this particular problem for 12 years, and they are now rolling out the first commercially available laundry-folding robot. They will be first to the global market and have secured a production partnership with Panasonic.

Shin and I talk a lot about AI and innovation in Japan, and also cover his rather unusual corse to innovation here. Seven Dreamers is not your typical venture-backed startup, and they might just provide a blueprint for innovation that many existing Japanese firms can follow.

It’s a great interview, and I think you’ll enjoy it.

Show Notes for Startups

Why AI can drive a car but not fold socks Why starting a company in Japan is different today Shin’s formula for developing innovative products How to work with large Japanese companies Why the future of laundry is more disrupting than you imagine Why big data wants to hack your washing machine The need to go global quickly Can Japan once again lead the world in AI

Links from the Founder

Everything you ever wanted to know about Laundroid Friend Shin on Facebook Seven Dreamers Homepage Find out more about Laundroid on Facebook or Twitter Nastent website Find out more about Nastent on Facebook or Twitter The carbon-fiber golf shafts on the Web and on Facebook

[shareaholic app="share_buttons" id="7994466"] Leave a comment

Transcript from Japan

Disrupting Japan, episode 81.

Welcome to Disrupting Japan, straight talk from Japan’s most successful entrepreneurs. I'm Tim Romero and thanks for joining me.

You know, the term artificial intelligence is thrown around far too loosely these days. Every start-up using decision trees, Bayesian algorithms, or the simplest machine learning techniques, label themselves as world leaders in AI. Now there’s no question that projects like Google’s driverless cars and IBM’s Watson have pushed the limits of what’s possible, and have introduced astounding innovations in AI over the past few years. But sometimes it’s surprising to take a look at the kinds of problems that are extremely difficult for AI. It turns out that folding laundry is one of those problems.

Today we sit down with Shin Sakane, CEO of Seven Dreamers and inventor of the Laundroid. The first commercially available fully automatic laundry folding robot. We talk a lot about AI in general. And the importance and the risk of attacking the really hard problems. And what he and his firm had to go through to make Laundroid a reality. It’s also worth noting that Seven Dreamers is not your typical venture back start-up. And Shin and I talk a lot about the role that mid-size companies have to play in kick-starting the Japanese economy and returning Japan to the global leader in innovation she was in the 60s and 70s. But you know, Shin tells that story better than I can. So let’s hear from our sponsor and get right to the interview.

[pro_ad_display_adzone id="1404" info_text="Sponsored by" font_color="grey" ]

[Interview]

Tim: So I’m sitting here with Shin Sakane of Seven Dreamers, and we’ve been bumping into each other for a long time now.

Shin: Right.

Tim: So thanks for finally making time and sitting down with me.

Shin: Thank you very much for coming.

Tim: We’re here to talk a lot about the Laundroid. Now it’s a robot that folds clothes, which I guess is the simple way of explaining it, but why don’t you tell us more about what it is.

Shin: Okay. We’ve been working on this project for the last 12 years almost.

Tim: Wow.

Shin: Yeah. We started back in 2005. This robot really folds daily life clothes and separate by categories or by family members.

Tim: Oh, okay.

Shin: Yeah.

Tim: So it can put all the shirts in one pile, or it can put all of dad’s clothes in one pile, and mom’s in the other—

Shin: Well, actually the key feature of Laundroid is that after you dry your clothing, you just put up to say around 30 clothes into insert box at the bottom. Then artificial intelligence and robotics with vision analysis technology, robot arms pick one clothing by one, and then it folds and put in the pick up box. If you choose separate by clothing category mode, then Laundroid puts towels in the towel tray, and t-shirts in the t-shirt tray.

Tim: Okay, and since this is a audio podcast, I guess we have to— The Laundroid, it’s a large machine, it’s about— What, two meters tall?

Shin: Yeah, about that.

Tim: And about 75 centimeters squared?

Shin: Right now it’s a little thinner. About 60 centimeters deep. Yes. And then 87 centimeters wide.

Tim: Okay.

Shin: So it’s a little thinner and wider.

Tim: And it is literally a black box.

Shin: Yeah.

Tim: It kind of reminds me of HAL in a way.

Shin: Yeah.

Tim: Well, it does have— It’s this big black monolith.

Shin: Right. Exactly.

Tim: With a bright circle on the front of it.

Shin: Right.

Tim: So you simply put up to 30 articles of what clothes in the bottom?

Shin: Right.

Tim: And then it sorts it on the shelves in the middle?

Shin: Right. Exactly.

Tim: Excellent.

Shin: And then if you want to separate by family members mode, in order to use this separation mode, you have to do very simple one-time registration process for each family member. If I purchase Laundroid, the first day what I do is put everything into the box. Just myself— my clothes. And then with your smart phone, register as a father, “father’s clothes”.

Tim: Oh, okay.

Shin: Yeah, robot arm put up one clothes by one, and then it takes so many pictures of each item.

Tim: So it learns by example.

Shin: Yeah, exactly. And it does automatically. It can remember the features of each clothing.

Tim: The field of AI is fascinating from the inside, but it’s very interesting from the outside in that— So you guys have been working on this for 12 years—

Shin: Yeah.

Tim: —to get to the point where you’re ready for production?

Shin: Right.

Tim: Why is this a hard problem? Why does AI have trouble folding clothes?

Shin: It was very hard, but not that hard to develop a robot to just fold, for example, t-shirt, towels, and pants. It took us about 3 years about to achieve that. But the condition is that we have to place say t-shirt, pants at a certain place first, then robot automatically folds.

Tim: Oh, okay.

Shin: That wasn’t that hard. That was hard, but that’s not that hard. The hardest part was just dump random clothing in the box, and then robot pick up one shirt or one clothing, and then reorganizes if this is t-shirt, or pants, or towels. And then reorganizes it if it’s upside down, or flipped, or something like that, and then place in a certain location in order to start folding. That’s the hardest part.

Tim: Okay, it’s not the folding that’s difficult. It’s the getting ready for the folding that’s—

Shin: Right, right. It’s so easy for human beings, but it’s so hard for robots and artificial intelligence.

Tim: So for example, there is a group in Berkley who’s built a robot that all it does is fold towels.

Shin: Yes.

Tim: Towels have to be the simplest thing possible to fold.

Shin: Right.

Tim: They’re all the same shape. There’s no real upside down to them. And it takes about a minute and a half per towel.

Shin: Right.

Tim: It does seem like folding clothes is one of those difficult problems.

Shin: Well, yeah it is difficult. But there are already three organizations who tried it, who achieved doing this rather than Seven Dreamers. One is as you mentioned, Berkley. Also like Berkley, one of the Berkley group achieved to also fold t-shirt just by using two industrial robot arms. But even when they do it, they have to place t-shirt in certain place first, then they start folding, right? So they achieved that, and you can see that on YouTube. And we think it’s worthless because—

Tim: Right, right. By the time you take—

Shin: Right, right.

Tim: The time it takes to position the towel.

Shin: Right, right. I might as well just fold it after for additional 5 seconds. So there’s another organization or company called Foldmate which is a US based venture start-up company I think based in Silicon Valley. Their machine— or from the CG picture— I don’t know if it’s real or not— but also customer has to place t-shirt or towel in a certain place.

Tim: Yeah, I have seen that, you have to clip it in—

Shin: Yes, exactly.

Tim: —the machine, right?

Shin: That’s what it is. Yes. So that’s another thing that we had that technology already back in 2008 and we didn’t commercialize because we thought no one is going to buy it because it’s really hard to do it. You know?

Tim: Right.

Shin: It’s too much trouble— hassle doing it. The third organization is the University of Tokyo. They’re the one who tried to do something similar to what we’ve done. But they could not also pick up the clothing and place from the random—

Tim: Okay.

[pro_ad_display_adzone id="1653" info_text="Sponsored by" font_color="grey” ]

Shin: —so they gave up.

Tim: So that’s the hard part.

Shin: Yeah, that’s the hard part.

Tim: And so Laundroid can fold anything? Pants, and shirts, and socks, and everything?

Shin: Yeah, pretty much all the regular clothes including t-shirt, and long sleeved shirt, and pants, shorts, and towels. And there are three things Laundroid can not do. One is Laundroid can not flip clothes. Meaning that the t-shirt has to be—

Tim: If it’s inside out?

Shin: Right, right. Exactly.