00:05:13 · Katie Prescott, Brennand Pierce
[00:05:13] S4: Tell us about that little robot. [00:05:14] S9: Yeah. So they've seen... It's interesting. Back in the day, they were like, what was called like Robo One in Japan. It was like 40¢ high fighting robots, and they were just made out of like survey motors you would find in a radio controlled car. We'd have to like modify them, make them actually do what you want and write all the software, make them balance and yeah, it was very, very cool and got me into it. And I spent, I think about two years, effectively on my own in forums, building it in in my my house in Bristol. Two years later, my house is now looking like a robotic lab, and I wanted bigger and bigger equipment. And quite coincidentally, at the same time, Bristol opened the first robotics lab, Bristol robotics lab, and actually did a master on like, alright, I am going to sell my first SAS company, and I'm gonna go and spend a year building robots at Bristol Robotics Lab. And it was a great year.
00:06:03 · Danny Fortson, Brennand Pierce
[00:06:03] S8: I'd love to get your take on what's happening right now because we talk about it all the time on the pod. I'm always like, if if ever there's a humanoid robot story or robot story, I'm always like, let's do that. And Katie always makes fun of me. You have all of the hype and all of the YouTube videos. Oh, here's a humanoid making a bed. Here it is cracking an egg. Here it is running faster than Usain Bolt, etcetera. What is actually happening, and how should we think about this moment, especially when you have, like, NVIDIA, Jensen Huang being like, yeah. We're putting a ton of money into robotics. We're trying to create a universal robot brain so that we can really hit the accelerator on this. Like, what is actually happening? [00:06:44] S9: I think it's obviously the next wave in technology. I was in Silicon Valley last week, and I I I got an Uber, Like, fifteen years ago, I went in my first autonomous car. You're like, this is the future. This is great. My friend founded the company, and you're in the back of the car driving around to Palo Alto. But it's taken fifteen years to go from that very early startup mentality to now I can order it with with an app, and it's quite mind blowing. And at the same time, and then my robotic... [00:07:12] S8: Are you talking about a Waymo?
00:07:13 · Brennand Pierce
[00:07:13] S9: Talking about Waymo. Yes. Exactly. And, you know, you're in the back of these cars going, like, 50 miles an hour with my other robotics friends. Like, so how much do you trust the programmers in your life, basically, on this? But at the same time, it gets you to the destination. It is it is literally like living in the future. And I think that is the same analog I see now with this robotics. Humanoid robotics is AI and the foundational models. I think it's where we were in autonomous car, like, ten, fifteen years ago, where it's like, the technology's there. It's just very immature. So I think there is a lot of not really smoke and mirrors, but it's staged, really expensive stage videos where you're like, alright. This is like proper production. It's doing a back flip, it's really amazing. But then at the same time, where are the deployments? If you could actually make a bed and do what all these startups could do, then you'd be selling, as you said, like, hundreds of thousands a year. I still think you're probably a couple of years away from really starting that ramp up where at the moment, most companies are just in the pilot phase. Most of them, I think, including my company, Bear Robotics, is very much looking at the low hanging fruits. Can you just work in a warehouse, pick up a tote, drive around, put it down? No one could really do that at scale yet, and that's quite a lot simpler than can you come into a house and do all my laundry and fold the sheets and everything like that. So I think it's like this incrementalism. But at the same time, I think just with Moore's Law and the acceleration of everything is is probably going to condense a lot of the timelines. It's probably not fifteen years like the autonomous car. It's probably more like five years until you start getting the early adopters, ten years until it's, like, everywhere.
00:08:52 · Katie Prescott, Brennand Pierce
[00:08:52] S4: But when you talk about early adopters, you you literally mean somebody having a robot in their house, folding the sheet? I mean, what what where are we on early adopters? What does that mean? [00:09:00] S9: Yeah. I think the home is is pie in the sky. You know? I see a lot of these startups in Silicon Valley is like, we're gonna do home robotics. You can't even, you know, work in a kitchen and do do dishes in a in, like, a Denny's or something like that, let alone actually in anyone's house. I think it's just like the incrementalism would be like, work in a factory is very structured. Can you do subassembly? So go go to a shelf, take the parts off, do the subassembly, put it in a box, and very repeatable. This is where we're doing pilots at the moment because it's... It takes a lot of time, like, a couple of months to set this up so it's actually working very reliably. We're getting up to, like, 98, 99% accuracy and repeatability, but that isn't what an actual factory worker or company wants. They want 99.99 uptises. It's trying to get that percentage up so it's actually doing useful work. And I think you've gotta have to go into the factory, in the warehouse, these sorts of tasks before you can actually start going out into more real world environments, let alone actually doing this task around people and being safe.
00:09:00 · Brennand Pierce, Katie Prescott
[00:09:00] S9: Yeah. I think the home is is pie in the sky. You know? I see a lot of these startups in Silicon Valley is like, we're gonna do home robotics. You can't even, you know, work in a kitchen and do do dishes in a in, like, a Denny's or something like that, let alone actually in anyone's house. I think it's just like the incrementalism would be like, work in a factory is very structured. Can you do subassembly? So go go to a shelf, take the parts off, do the subassembly, put it in a box, and very repeatable. This is where we're doing pilots at the moment because it's... It takes a lot of time, like, a couple of months to set this up so it's actually working very reliably. We're getting up to, like, 98, 99% accuracy and repeatability, but that isn't what an actual factory worker or company wants. They want 99.99 uptises. It's trying to get that percentage up so it's actually doing useful work. And I think you've gotta have to go into the factory, in the warehouse, these sorts of tasks before you can actually start going out into more real world environments, let alone actually doing this task around people and being safe. [00:10:05] S4: So that sort of work is is five years away, but still we have advanced so much from where you started out getting those deliveries from Japan building your little mini robot. Just tell us about that journey and Bear Robotics and and what you're doing now. How did you go from those mini robots to considering putting your robots in factories?
00:10:05 · Katie Prescott, Brennand Pierce, Danny Fortson
[00:10:05] S4: So that sort of work is is five years away, but still we have advanced so much from where you started out getting those deliveries from Japan building your little mini robot. Just tell us about that journey and Bear Robotics and and what you're doing now. How did you go from those mini robots to considering putting your robots in factories? [00:10:26] S9: Yeah. Because when I finished my master's in in robotics, I looked out into the field, wanted the job, didn't want to stay in academia. But at the time, it's just like factory arms. It was very boring. So I spent the next, like, I eight years in academia at Carnegie Mellon, Technical University of Munich, really working on humanoid robots, getting them to just walk and, you know, balance. And some people even trying to get them just to stand on a skateboard was was bleeding edge like ten years ago. About ten years ago, I looked at the field. I wanted to go back and do another startup. There was a lot of people I knew setting up like Fetch Robotics and a lot of these companies. I would have loved to do a humanoid robot, but back then, it's very impractical. So I went to use a lot of my research in, like, navigation, SLAM, things like that, and actually founded Bear Robotics, which was doing delivery in restaurants. So that was just doing a very mundane, repeatable task. You take the food out of the kitchen, put it on the robot, it takes it to the table. We've been doing that now about nine years, shipped like 16,000 robots, and [00:11:28] S8: doing very well. Where?