From Childhood Coding to Construction Innovation: Tessa Lau's Robotics Revolution 的完整文字稿

00:02:46.276 I wanted to be an author for a while because I really love writing and I just never imagined I would be a CEO or run a technology company. 00:02:46.276 --> 00:02:54.025 It was just like so far out of the realm of what little kids like me could do that I just never imagined anything like this. 00:02:54.025 --> 00:03:00.794 Well, I can see the autobiography coming in another 10 years or so. 00:03:00.794 --> 00:03:09.014 I've got a lot of stories built up so eventually they'll come out Well, good, good. 00:03:09.034 --> 00:03:11.387 Were there any specific events or influences during your childhood that set you on this path and into technology? 00:03:11.387 --> 00:03:14.596 And you know you mentioned the, you mentioned the computer that your dad gave you. 00:03:14.596 --> 00:03:16.781 Any other specific milestones or events? 00:03:18.751 --> 00:03:26.781 I mean, the only other thing I remember really is is just like, uh, what got me into technology and I think a lot of kids too is video games. 00:03:26.781 --> 00:03:33.878 Uh, and I don't think people appreciate the impact that gaming has on the technologists that we have today. 00:03:33.878 --> 00:03:42.947 Right, oftentimes that's people's first uh touch or first way to interact with anything technologically cool. 00:03:42.947 --> 00:03:59.219 Video games always push the envelope of what's possible, and so I remember, like, having an 8-bit nintendo back in the day that I just thought was the coolest thing, right, um, because all of these games that came out for it were just like, so advanced for the time. 00:03:59.219 --> 00:04:03.192 And that's what got me into programming, because I wanted to build my own. 00:04:03.192 --> 00:04:07.203 And so I think a lot of kids probably share that same path. 00:04:07.203 --> 00:04:15.102 You know, they want to create something cool, they want to tell a story, they want to build something that no one's ever seen before, and that's what got me in for sure. 00:04:15.829 --> 00:04:16.191 Very cool. 00:04:16.191 --> 00:04:17.153 Yeah, certainly for me too. 00:04:17.153 --> 00:04:25.625 I can relate Video games is what drove me into building computers and learning basic programming and scripting and eventually you know website development and everything was all because of. 00:04:25.625 --> 00:04:27.898 You know video games and the community that was built around it. 00:04:27.898 --> 00:04:32.959 Yeah, that was a huge part of my childhood and eventually graduated out of it myself. 00:04:32.959 --> 00:04:43.043 But you know, I appreciate so much and now I feel behind the times, right, I mean, like you said, video games are always pushing the edge of technology and moving it forward. 00:04:43.043 --> 00:04:45.078 So, yeah, I'm proud of everybody in that space. 00:04:45.078 --> 00:04:54.441 Well, so you've grown up since then and you've had a remarkable career journey, and you know you've been at, as mentioned, ibm Research and Willow Garage. 00:04:54.441

--> 00:04:59.862 What are some particularly challenging moments at either of those two spaces and how did you overcome it? 00:05:01.838 --> 00:05:04.950 challenging moments at either of those two spaces, and how did you overcome it? 00:05:04.950 --> 00:05:06.111 I think one of the. 00:05:06.111 --> 00:05:10.382 I want to tell the story about how I get into robotics, because that kind of like blends the how I got to where I am with some of the challenges. 00:05:10.382 --> 00:05:14.500 So my first job out of school was at IBM research. 00:05:14.500 --> 00:05:29.403 I spent 11 years there doing research in human computer interaction and and AI and and after what felt like a whole career at IBM, I wanted to do something different and so I decided to get into robotics. 00:05:29.403 --> 00:05:45.877 And the reason I got into robotics as opposed to anything else was because I heard about this project at Willow Garage that was called Robots for Humanity and the intent of the project was to work with this person with disabilities. 00:05:45.877 --> 00:06:00.579 He was a quadriplegic, bound to a wheelchair, couldn't move his hands or anything other than a little bit of a finger and his eyes, and they were allowing him to program a robot to help himself with activities of daily living. 00:06:00.579 --> 00:06:09.153 And that kind of blew my mind, because I've worked in software pretty much my entire career up to that point and all of the things that I built. 00:06:09.153 --> 00:06:19.893 I built a lot of things, but they were living behind a screen, they couldn't touch the world, and so that really inspired me to get into robotics, because I realized that robots can actually touch the world. 00:06:19.893 --> 00:06:24.939 They can actually make things happen in reality, not just virtually. 00:06:24.939 --> 00:06:27.263 And so I got into robotics. 00:06:28.004 --> 00:06:43.571 It was a really tough learning curve Like there's a lot to learn about robotics, and I kind of had to like start from scratch and learn a lot of things you know from, from the basics, but it, you know, led me to the point where I am today, and so it's definitely worth it. 00:06:43.571 --> 00:06:48.002 But it's overcoming that feeling of like you're starting from zero again. 00:06:48.002 --> 00:06:54.111 You're starting as like a complete noob in a new area, a new domain where everyone else knows a lot more than you do. 00:06:54.111 --> 00:06:56.838 There's a lot of imposter syndrome in there. 00:06:56.838 --> 00:07:04.165 There's a lot of fear that you're not going to be able to catch up, but when you push through it you can actually make some really interesting progress. 00:07:05.028 --> 00:07:07.114 And congratulations on your perseverance as well. 00:07:07.114 --> 00:07:17.751 I'm thinking about the PhD component to what you have, and you know that's just an envelope of saying you know, hey, I can take uncharted territories and learn so much about it. 00:07:17.751 --> 00:07:19.877 So congratulations on that, tessa. 00:07:20.338 --> 00:07:20.559 Thank you. 00:07:24.210 --> 00:07:36.615 So prior to Dusty Robotics, you were also the CTO at Savvy Oak, where you deployed over 75 delivery robots in the hospitality industry, so I'd love to learn more about the experience there and also what was the most rewarding part of that experience for you. 00:07:38.451 --> 00:07:41.100 So Savvy Oak came out of Willow Garage. 00:07:41.100 --> 00:07:59.314 So, as Willow was shutting down, so Willow was the premier research lab for robotics at the time and they ran out of money, decided to shut down, and a number of us at Willow decided to start a new company, savvy Oak, to commercialize some of the technology that we developed at that research lab. 00:07:59.314 --> 00:08:09.361 And so we settled on hotel delivery, because we've all stayed in hotels you all need to, you know, have left your toothbrush at home and you need a little emergency delivery in the middle of the night. 00:08:09.361 --> 00:08:17.816 And so we thought that that would be a good thing to build, because it would make use of the navigation capabilities that that we were developing for robots at that time. 00:08:17.816 --> 00:08:26.611 And you know, with our robotics arrogance, we thought, with our, uh, robotics arrogance, we thought, oh yeah, navigation is a solved problem, so it should be really easy to build this robot, right? 00:08:26.611 --> 00:08:28.374 Uh, yeah, no. 00:08:28.374 --> 00:08:31.720 So, um, we, uh. 00:08:31.720 --> 00:08:46.240 One of the biggest things that I learned there is that when you try the first time, you try to bring a technology out into the world, the first time you try to bring a technology out into the world, the hard parts come when you actually get the 20% at the end, right? 00:08:46.240 --> 00:08:48.764 So 80% of it's, you know, pretty straightforward. 00:08:48.764 --> 00:08:52.307 It's like the middle of the road use case it's that 20% that can kill you. 00:08:57.834 --> 00:09:01.110 And so for us at Savio, that 20% was dealing with the elevators. 00:09:02.149 --> 00:09:06.498 Because if imagine you're in a hotel and you need to get from the front desk up to the room, typically that revolves an elevator trip, right? 00:09:06.519 --> 00:09:18.512 And so one of the things I'm most proud of at Savioke was doing the human robot elevator interaction programming a robot to take the elevator like a human does. 00:09:18.512 --> 00:09:21.139 And if you think about it, it's kind of complex, right? 00:09:21.139 --> 00:09:37.617 When you walk into an elevator, you turn around to face the door, so in order for our robots to and you have to sometimes wait for other people to get off before you can get off, or if you're in the front and someone behind you needs to get out, right, maybe you have to step off the elevator and get back on. 00:09:37.617 --> 00:09:39.692 We've all been in these situations, right? 00:09:39.692 --> 00:09:47.796 And so, if you

think about how complex those situations are, it's really hard to program a robot to know what to do in all of those different situations. 00:09:47.796 --> 00:09:59.844 And so that was, you know, watching our robots take elevators and behave politely, despite the amount that guests have been drinking on that elevator. 00:09:59.844 --> 00:10:04.855 That's been a real, that was a real success for us, I think, and I was really proud of that. 00:10:05.738 --> 00:10:07.730 That's fantastic, and so where did that end up going? 00:10:09.473 --> 00:10:18.972 So Savio, rebranded as Relay Robotics, still going, and so I left after five years because I wanted to pursue other dreams. 00:10:18.972 --> 00:10:20.615 But I wish them the best of luck. 00:10:20.615 --> 00:10:22.337 They're still plugging away at it. 00:10:23.058 --> 00:10:23.519 Very cool. 00:10:23.519 --> 00:10:24.661 Thanks for sharing that story. 00:10:24.661 --> 00:10:31.919 So starting Dusty Robotics from scratch, securing funding and building a great brand and reputation is no small feat. 00:10:31.919 --> 00:10:39.692 Since you were last on our show in season one, what were some of the biggest obstacles you've encountered at Dusty Robotics and how have you navigated them, Tessa? 00:10:40.293 --> 00:10:49.613 Yeah, so I was on your show a couple of years ago and in a startup, each month feels like a year, so it feels like a really long time ago. 00:10:49.613 --> 00:10:59.358 We've made some amazing progress since then, but also solved a number of really significant challenges. 00:10:59.358 --> 00:11:06.721 I think the thing I'm most proud of of having worked through is hiring a really, really good team. 00:11:06.721 --> 00:11:10.279 I think the team is what makes the startup. 00:11:10.279 --> 00:11:21.379 It's what gets me out of bed every day to come to work and work with this group, and my team is the one that's actually doing most of the work and actually solving those hard problems for me. 00:11:22.250 --> 00:11:27.096 One of the things I realized is that, as CEO, my job becomes less and less Over time. 00:11:27.096 --> 00:11:45.607 It becomes less doing and it becomes more judgment and vision and managing, and so right now, the hard parts of my job is making sure I have the right people in the right roles to do the right things for the company, and so that's an ongoing set of challenges. 00:11:45.607 --> 00:12:06.862 We're constantly looking for good talent and defining what we want them to do and bringing them in, and so I would say some of the biggest challenges I've had are, essentially, designing the organization to accomplish the goals that I have for the company and then finding the right people to fill those roles inside the organization, and I'm really, really happy with what we've been able to accomplish there. 00:12:08.086 --> 00:12:09.087 And likely. 00:12:09.087 --> 00:12:12.216 I'm proud of you, tessa, and I'm proud of your team as well. 00:12:12.216 --> 00:12:18.951 You know what the impact I've seen you make in the industry over the last two years from the outside in has been astronomical. 00:12:18.951 --> 00:12:29.992 You know it's been amazing for me to sit in SMACNA association meetings and then talk about the training that's about to be delivered to union members, about Dusty Robotics and that there's. 00:12:29.992 --> 00:12:40.872 The speed at which you've been able to move in this industry and help bring positive momentum and positive change in this industry has been really awe-inspiring from the outside in. 00:12:40.872 --> 00:12:45.048 So I just want to say how proud I am of you and your team for the progress that they've made. 00:12:45.589 --> 00:12:48.637 Yeah, well, I'd love to give a shout out to our head of training, marshall Hawley. 00:12:48.637 --> 00:13:10.512 So a lot of that work has been his idea and his doing, and so I'm super happy to say that we're now working with several unions, not just in California but all across the West Coast at this point, and Marshall has created a fantastic curriculum for them to use to teach to their membership, and that's part of the way we're scaling this up. 00:13:10.512 --> 00:13:26.567 We can't personally touch every single person who's going to use and operate our system, and so the way for us to scale ourselves is to work through the unions, get their trainers on board to teach our product, and Marshall's been fantastic at getting all of that off the ground and going. 00:13:27.208 --> 00:13:28.993 Yeah, so cool to see from the outside in Tessa. 00:13:28.993 --> 00:13:31.748 It's just amazing, yeah, and well done Marshall as well. 00:13:31.748 --> 00:13:39.996 So, speaking of the construction industry, it definitely has a perception of being relatively slow to adopt new technologies. 00:13:39.996 --> 00:13:46.933 What do you see as the biggest barriers to technological adoption and construction, and how is Dusty Robotics addressing and breaking down those barriers? 00:13:48.358 --> 00:13:51.947 I think the biggest barrier is that stuff just doesn't work. 00:13:51.947 --> 00:13:55.831 Let me expand a little bit on that. 00:14:01.235 --> 00:14:13.150 So, building a product is hard, and building a product that people want is even harder, and if you can manage to do both of those, then you can win, and that's what we've done. 00:14:14.206 --> 00:14:27.868 And I think the reason why construction tends to be very suspicious of new technology is because they've probably been promised a lot of things that don't actually pan out. 00:14:27.868 --> 00:14:47.062 The products don't actually live up to the promises, or maybe the product doesn't actually do what construction companies need it to do, and so I think one of the reasons we've gotten so far as we have is because we listen really closely to what our customers need. 00:14:47.062 --> 00:14:49.168 We really understand their pain points. 00:14:49.168 --> 00:15:06.498 We hire from the construction industry, so we

have a lot of that expertise in-house and we have a philosophy of building something that's 10x better than what the industry is doing today, and if it's not 10x better, then it's not meeting our goals for that product. 00:15:06.498 --> 00:15:18.995 So what that means is that anything that you get from Dusty is guaranteed to give you a 10x improvement, and for most of our customers that means 10x speed, accuracy and much better communication as well. 00:15:20.257 --> 00:15:30.928 Very exciting, yeah, and I hear what you're saying when you're saying you know, hey, is the product of need, Is it exactly what I want? 00:15:30.928 --> 00:15:43.165 And then is it going to perform and deliver to the expectations that I have or that have been set for an environment that's mission critical right, and things need to work. 00:15:43.165 --> 00:16:02.778 Things need to produce mission critical right and things need to work, things need to produce. 00:16:02.778 --> 00:16:15.471 So, in your opinion, speaking about other technologies in the AEC or architecture, engineering and construction space, what do you see outside of robotics that appears most promising and more trending towards? 00:16:20.694 --> 00:16:21.736 you know, towards adoption in the next few years. 00:16:21.736 --> 00:16:42.557 Yeah, and I just want to underscore what you just said, because that is so true, right, part of the reason why it's hard to get new technology into construction is because the risk profile in construction projects is just so painful in the layout business, right? 00:16:42.557 --> 00:16:47.097 So I hear a lot of layout stories and I've heard stories about construction companies that go out of business because they made a mistake in layout, right, and that's a big deal. 00:16:47.097 --> 00:16:53.321 If you can kill your company because of just this tiny little thing, that's huge. 00:16:53.321 --> 00:17:00.617 And so people are very risk averse, right, because a mistake can cost them not just their job, but it could cost them their company. 00:17:01.719 --> 00:17:11.548 So I totally get why people are not willing to take as many risks in construction, and so that's why our philosophy is trying to minimize that risk, right. 00:17:11.548 --> 00:17:17.302 In fact, we're giving you technology that decreases your risk it doesn't increase it, right. 00:17:17.302 --> 00:17:22.201 So that's just how I think about construction technology adoption. 00:17:22.201 --> 00:17:26.086 So you asked what other technology is coming down the line. 00:17:26.086 --> 00:17:32.806 Everyone's talking about AI these days, and I'm sure that's actually going to make a huge impact in construction. 00:17:32.806 --> 00:17:40.115 I'm actually tracking some really interesting use cases applications for it, the two that I'm really excited about right now. 00:17:40.115 --> 00:17:44.158 One of them is generative AI for modeling. 00:17:44.939 --> 00:18:10.788 And so Hypar and TestFit are the two ones that I'm super interested in there, because they're basically helping architects and designers and detailers create better designs by using AI to generate the starting point for those designs, in some cases, maybe even like a full-on, fully populated out 3D design, and so that's going to save a lot of time. 00:18:10.788 --> 00:18:14.976 It's going to eliminate a lot of errors that happen during the design phase, so I'm really excited about that. 00:18:14.976 --> 00:18:25.059 And then the second area that I'm really excited about is companies like Trunk Tools that are using AI to create a conversational interface to your construction documents. 00:18:25.059 --> 00:18:41.502 So, instead of your superintendent having to pull up Procore and scroll through a lot of different documents in order to find the information they need, they can just ask a question of like a chat GPT type AI, and it searches through all those documents for you, gives you the right answer, saves you a huge amount of time. 00:18:46.634 --> 00:18:47.357 Very cool. 00:18:47.357 --> 00:18:51.214 Yeah, thanks for sharing those insights and also specific companies to look after. 00:19:01.241 --> 00:19:07.790 Looking ahead and coming back into robotics, what are your to ask the question of what robots are you using? 00:19:07.790 --> 00:19:10.724 Because we don't ask do your tools use electricity? 00:19:10.724 --> 00:19:20.167 And it's kind of taken for granted that they do, and it becomes a foundational technology that just gets baked into every single product. 00:19:20.167 --> 00:19:27.884 The same way, there's a battery and electrical power on your power tool right, there's going to be robotics embedded in your power tool. 00:19:27.884 --> 00:19:37.322 And so the way I see that playing out like, if I think about my life right now, you know I drive a Tesla, which is a big self-driving robot. 00:19:37.322 --> 00:20:00.416 It's like a car, but it's got sensors and it can take a lot of decisions on its own and gets better over time and it's on track to becoming fully autonomous for this one particular use case right, I'm not going to get it to build a house for me, but for that one use case that it's designed for, it's doing that really, really well and it's using robotic technology in order to do that. 00:20:01.038 --> 00:20:02.764 Another example is my Nest thermostat. 00:20:02.764 --> 00:20:42.462 It's got sensors and knows how warm the house is, knows when it needs to turn on the heat or turn on the AC, and it acts in response in order to deliver a certain outcome that I want it to, and so those are all characteristics of robotics that are using technology that was originally developed in the robotics field, and we're starting to see those in everyday products today, and so my future of robotics is that you're not gonna see necessarily a humanoid robot wandering around doing all these things, but the technology that you already use on a daily basis is going to start incorporating robotic

features in it, and they're going to be little robots. 00:20:44.289 --> 00:20:44.670 Very cool. 00:20:44.670 --> 00:20:46.673 Thank you for sharing that vision of the future. 00:20:46.673 --> 00:20:49.655 I think of my mind also goes to sort of nanorobots. 00:20:49.655 --> 00:20:51.437 Where are your thoughts on that? 00:20:51.437 --> 00:21:00.165 I was just reading an article that that could sort of you know, nanorobots could help organs, you know, stay healthy and have humans live for a long time. 00:21:00.666 --> 00:21:00.807 Yeah. 00:21:01.191 --> 00:21:03.338 Do you see a correlation any of that, or is that too far away? 00:21:04.110 --> 00:21:08.000 I admit I find the concept of having little robots like crawling around in my bloodstream. 00:21:08.000 --> 00:21:09.102 I find that kind of creepy. 00:21:11.572 --> 00:21:12.336 Yeah, you're not the only one. 00:21:12.336 --> 00:21:15.740 The author of the article said something very similar. 00:21:15.740 --> 00:21:17.135 It's like is that going to be accepted? 00:21:17.135 --> 00:21:26.574 But yeah, I'm very curious about the world of nanorobots and how far away humanity might be from that concept. 00:21:27.409 --> 00:21:29.935 Well, I am excited about the surgical robots. 00:21:29.935 --> 00:21:34.573 One of my friends just had heart surgery right, and they don't open your heart anymore. 00:21:34.573 --> 00:21:48.165 They they send a little device up, you know, into your artery, through your arteries, in order to do whatever it is they need to do on the inside, and that's like so much easier to heal from right. 00:21:48.165 --> 00:22:05.131 It's much less invasive and that's possible because of the precision of those little robotic devices that can more autonomously perform certain actions inside the body, and so that's exciting yeah I'll get sick at some point. 00:22:05.151 --> 00:22:17.664 We're all going to need medical care, and the better that is, the better we all are absolutely and for them to not have to do the open heart, the open, open chest portion of the of the surgery I mean that you know, probably saves months of recovery. 00:22:17.664 --> 00:22:29.670 Um well, how do you, uh, how do you envision dusty robotics evolving over the next few years, and at what impact do you do you feel like you will have, or hope to have, on the construction industry over the next few years? 00:22:30.813 --> 00:22:46.500 So I think Dusty is in a really interesting space, because where we've started, which is this problem of layout, is this kind of critical junction right now where the digital world meets the analog world of the field, and where I want Dusty to go. 00:22:46.500 --> 00:22:49.990 My vision for Dusty is because we've started in the middle. 00:22:49.990 --> 00:23:02.574 Here we have the opportunity to both go upstream of where layout happens, as well as downstream of where layout happens, and so if you think about what happens upstream, well, where does the file come from that our robots print? 00:23:02.574 --> 00:23:07.653 Right, that's an output of the design process, of the BIM coordination process. 00:23:07.653 --> 00:23:17.160 There's a lot of people working together in order to create that data that our robots print on the ground, and right now, all of those processes are still very manual. 00:23:17.160 --> 00:23:22.842 I still see people schlepping files around on a USB key or in email. 00:23:22.842 --> 00:23:29.798 I still see phone calls being made to coordinate and discuss things, and so how can we automate that? 00:23:31.151 --> 00:23:33.839 That is a process that we're looking into as a company right now. 00:23:33.839 --> 00:23:46.933 We're starting to build some tools that help streamline that BIM coordination process in order to get all the trades on board and all of the information lined up in the right way so that when you actually do go to print it. 00:23:46.933 --> 00:23:49.638 You're confident that that's exactly what you want to build. 00:23:49.638 --> 00:23:56.135 So that's one thing, and then, downstream, I'm also really excited about other use cases. 00:23:56.135 --> 00:23:59.990 So we have lots of ideas about other robots we could build. 00:23:59.990 --> 00:24:21.470 I'm not going to talk about any of them today, but if you think about it, the BIM model is the instruction manual for the building that you want to build, right, that tells you exactly what you want to build, and once you have those instructions which we do because of the nature of our job that is basically telling future robots what they need to do and where. 00:24:23.394 --> 00:24:23.813 Very cool. 00:24:23.813 --> 00:24:29.605 I'm so excited and I'm happy to see you expanding to, you know, the broader spectrum of the industry challenges. 00:24:29.605 --> 00:24:41.925 Well, tessa, as a final question of the show, and a traditional future construct, if you could project yourself out 25 years and wanted to have any device or technology that would benefit you personally, what would it be and what would it do? 00:24:42.829 --> 00:28:16.449 no-transcript.