Tune in for a conversation with Strategic Advisor and robotics expert, Ken Goldberg, PhD to learn more about the challenges and progression of robotic technology over the last few decades, and how it is changing our world today.
Good afternoon. I am Bill Studebaker, president and CIO of ROBO Global. And I'm honored to be here with you today to talk about trends inherent in robotics and artificial intelligence. I'm joined by Dr. Ken Goldberg, who is a ROBO global strategic advisor. Ken is also a professor and chair of industrial engineering at UC Berkeley. And Ken is a distinguished roboticist and entrepreneur, that holds dual degrees in electrical engineering and economics from UPenn and a master's and PhD from Carnegie Mellon. Ken joined the faculty of UC Berkeley in '95, and he's been researching robotics for nearly four decades. So he has a pretty unique perspective. And after 20 years of researching robot manipulation, grasping, Ken co-founded Ambi Robotics, which is a universal bin picking robot that has the ability to do superhuman sorting at twice the speed of manual picking. So today, Ken, welcome.
Thank you. Thank you, Bill. It's a pleasure to be here. Thank you for that nice intro.
Thanks for coming. So today we're going to talk about the trends, again, inherit in automation and just the tremendous progress that we're seeing and discuss areas of growth, as well as challenges. And piggybacking on this, I do want to comment that the research team at ROBO Global just completed our annual trends report for 2023, which we hope you find pretty interesting, as it should illustrate our conviction in the robotics and AI investment opportunity. As sort of a prelude to our conversation, I would like to say that we expect to see technology and innovation solve problems, as it has throughout human history. And clearly, the digitization of the economy is proceeding at full speed. Fortunately, innovations on sale for investors, unless you feel that, or at least we do, I do at ROBO Global, that automation is not dead. We think it's a perfect time for investors to buy in this pullback, given that many innovation stocks are off 50-90%. Ken, I'm just curious on, given your domain expertise, that you could share your perspective on the technology and the progress, that we've seen over the last few decades, as well as some of the challenges. And I'd be curious to also get your insights on what industries are seeing faster adoptions than others and what are some of the technical hurdles that are hurting other industries. So with that, love to hear your thoughts.
Great. Well, thanks Bill. I have been saying that I see the period we're in as something like the roaring '20s of the of the last century. And that time, if you remember, they had just come out of their pandemic, the 1918 pandemic. And then there was this huge amount of exuberance and creativity and energy. Basically, everyone wanted to appreciate getting back together and getting out again. And so, I think we're in a very similar situation. I see a huge amount of enthusiasm, that is expressing in a variety of different directions. We also have, of course, our challenges economically with inflation, with the war. But I think that for robots specifically, there's certain sectors that are moving in a very exciting directions.
And the one I know best is logistics, as you mentioned. And this is also been affected by the pandemic, in that the demand for e-commerce has skyrocketed. And it's just change in behavior. People are just ordering things in a way they didn't three years ago. And that's happening at the consumer level. It's also happening at the business level. And the challenge is how do you keep up with that demand. And that means how do we get those products actually out to customers? And so there have been a lot of challenges. The supply chain is still getting resolved. But a big one is just in the shipping and getting huge numbers of packages out, especially when there's a lot of variation in the volume.
So there's a huge upswing. And what's been exciting from my perspective is that the robots are really being adopted now to assist in the management of logistics. So Amazon, for example, for years has been using robots in warehouses to facilitate moving shelves around. So these kind of automated vehicles are more and more adopted in many different warehouse contexts. But the next step is to actually be able to take things out of the shelves and out of bins and be able to pick them up. And that's the area that I've been working on. As you mentioned, I've been working on the same problem for 40 years. And I've made remarkably little progress. It's a hard problem. And I want to just give you a sense of why that is. I mean, people pick up things like this all the time, and they do this and it's very easy. Even a child baby can do that.
Now that seems so incredibly easy. It's much easier than playing chess, for example. But robots will still have an incredibly hard time picking this up, let alone doing this with it. And why is that? Well, it's very subtle. I can say that the more I study it, the more I appreciate the human ability. But it has to do with three aspects. There's uncertainty here in actually the perception, because it's very hard.... You see that this is transparent, and so it's very hard to actually make out where the edge of it is. And we do it very easily as humans. But robots and artificial systems have a hard time being able to see the edges of something transparent. So it's perception.
The second is control. So even if you knew where the edge was, getting your robot fingertip to the right spot is a challenge. And that's because of the inherit uncertainty in the gears and the motors and the control system. And then there's a third source, which is uncertainty in the friction and the physics. You have to know where the center of mass this thing should be and how basically slippery is it. And so all those things are uncertain. And so, a very small error in any one of them can cause the object to be dropped. So even a microscopic error can cause it to be dropped. So the challenge is, "How does that work? And how do we get robots to be able to do it well?"
And the good news is about 10 years ago there was a breakthrough in deep learning, and everyone knows that's the AI revolution. And it took some time, but we found one approach to using that, that surprisingly turned out to work remarkably well. And that is to train the system on many simulated objects and their geometries, and then it would generalize to new objects, that it had never seen before. And we'll share with you that the system called NExTNet was very successful. We published a bunch of papers, and it was covered in the press. One thing we always showed as an example of something you couldn't pick up was this. This is still basically extremely difficult to be able to pick up. We haven't solved everything. So there's a number of problems with things that are very hard to pick up. So the challenges are still there.
But progress has been made to the point where we spun out a company, and Jeff Mahler was the brilliant PhD student, who is joined by Steve McKinley, David Gilley and Matthew Matl. And so the five of us co-founded Ambi Robotics. And I would say they have been working especially hard on really building a commercial system. And they brought in a brilliant CEO, Jim Leifer, who really knows the business of the of logistics and warehouses. The company is up to 50 people. And we are producing systems called AmbiStore, that we've now installed in 70 facilities around the US. And these are sorting tens of thousands of packages as we speak. Particularly, it was a race to get all this set up before peak season. So the team spent all summer making this happen, and now the systems are up and running and reliably. And we're now just basically hunkering down to keep them all fine-tuned so that they'll get through the season. So this I'm very excited about. I think this can continue and this will expand. We have something like 1% of the market out there. And so there's a lot of room for expansion. And I'm very bullish about that area. I think that's an area that robotics has really matured, and it's a sweet spot, really, for robotics.
Ken, maybe you could just share with the viewers what makes Ambi a successful technology. Obviously, you've spent 20 some years on research, and it's been a lot of development, and you are beginning to solve a problem that's been inherently difficult with robots, which is to grasp unstructured items. It's easy for a robot to pick up a structured similar item, and it can do it quite easily. But it's a lot different when you have variations, and curious to understand your technology a little bit more.
Sure. Well, one of the things is that, as you said, the technology there, it's a variety of elements that were developed outside of the university. So the group left Berkeley and then started the commercialization process. So all the software has to be rewritten, has to be especially fast. It has to take into account not only a single, in this case, suction cup, but multiple suction cups. And you also have to worry about motion planning. And that means, once you grab the part from a bin, how do you get it out without colliding with the bin or other things? That turns out to be surprisingly subtle and complex. And doing that computation fast is another big challenge. You essentially have to be doing this at a pretty blinding speed, so that you can keep with the pace of these logistics centers. So it's advances in software, but also in the hardware.
And the team has discovered and invented a number of innovations, both in the tooling, in the mechanisms, that allow the system as a whole to work. So the system is about the size of a 18-wheel truck. The robot is one part of it, the robot arm, but then it has from up to 60 bins on the other end. So it picks up the part, scans it, drops it, and then it gets shunted down with a shuttle dropped into the bin. So all those interacting components have to work together. And you have to think about things like... And very important, when you said, "What is the secret?," if you will, I would say it's customer focus. That is the key. And it means that knowing who the customer is, really understanding what their needs are and concerns.
So one thing we've learned, and I think it's been very interesting, is that, as a technologist, I might think, "Hey, we've got this great technology. Let's come in and this is going to solve your problem." Well, turns out that the problem is different. The technology is only one part of it, but they want a whole system. And the whole system has to work and has to be interfaced. And you have to write manuals, and you have to fail-safes, so nobody gets hurt, and so when something does go wrong, that it doesn't break down the whole system. And there's a variety of things. And it has to be able to be installed fast. There's a variety of things that you don't think about. And so those ingredients are really part of the company and part of the DNA, which is we are really working alongside with the workers. From Berkeley, we're power to the people. We're very much on the I-side of getting things done.
And so workers actually like our machines. When they have a problem, they call us. And they say, "We want to fix this as soon as possible." So that's a good sign. We have really good relationships with the companies we're working with. And the technologies, I mean that's the other thing, Bill, that we've watched those evolve. And so the technology, that piece of AI that we're using, is now very reliable. And that is very exciting for us, that sim to real idea, that was a conjecture five years ago. Now it's really proving out, and it's working around the clock.
Okay. Well, kind of piggybacking on the comments of robots working alongside of people, there's been a lot of skeptics about automation, about robotics and AI, and a strong narrative that robots are stealing our jobs. I actually find that to be kind of an untrue statement. There's roughly going to be 4 million industrial robots installed globally by the end of next year. Put that in some comparison, there's roughly about 500 million people in manufacturing globally. There's a little less than 1 robot per 100 workers. So if robots are stealing our jobs, they're doing a bad job of it. And I think what's interesting about it, and you've talked about it, Ken, is that robots are pretty complex tools that really help amplify the human capability. And humans and robots really are best when collaborating. I'm just curious your perspective on this and how people should think about this.
Well, and thanks for asking. I think that is actually exactly right, Bill. The key is that robots are there, when they're designed well, these are machines that actually increase our productivity. So there are some cases where robots replace humans, of course. But the vast majority of cases is where you have systems that integrate and allow the overall manufacturing site, or the overall warehouse, to be much more efficient. So there's a big sense of progress there, and that workers, actually, they feel better about the job, because they're getting more throughput. They're being more productive as a group. And this has been seen over and over again. Unions used to be very opposed to automation. And they gradually came around to viewing automation as a benefit, because it meant that there was more investment in the different facilities and showed that those facilities were more successful when they had automation. So that actually meant job security for the workers.
So when we're talking about the workers in these warehouses, they're not going to lose their jobs. In fact, the hardest thing is to keep workers, because the turnover is really high. These jobs, there's a lot of injuries. People just burn out. But if you can make the job less stressful and onerous, then all of a sudden the work is better for the humans and more work gets done. So the key is thinking about the robot as a compliment, complimentary to the human workers. And the examples of that, they sometimes say, "Well, are we going to be putting journalists out of work?" Some people are claiming that. I don't think that's going to happen at all. What's going to happen is you're going to have tools that AI can help journalists focus on what is most important about their jobs. So transcribing a conversation like this isn't a good use of a journalist's time. They now can use tools like AI that we have in on Zoom and Google Translate to translate into another languages. These are all tools that can help the worker be more efficient. They don't replace the worker.
And the other example I like to use is, if you think about Uber and Lyft and Google Maps, Google Maps and the Uber and Lyft applications, they just make transportation so much better than it was at five years ago. It's because of those two things. It's now an app; it helps coordinate where people are. You can allocate effort, and you also don't have the problem of finding maps and getting lost. I realize that there used to be a pleasure in getting lost sometimes, and I hear you. But I would say for the most part, it was not a pleasure getting lost, and it was a hassle. And you had this map, and I remember how stressed out you would be trying to get somewhere. You're late, and you don't know where you are. That's pretty much gone. It's gone away, especially if you're a taxi driver or a truck driver.
So I think that the technologies we have to recognize are greatly enhancing the job, making us all more productive. And I think that is going to continue. And that's where I think ROBO Global is thinking about that from a really strategic position, is thinking about where are these advances and how are they going to improve the efficiency and productivity of these industries.
Well, it's interesting, Ken. I mean, I like to think of robotics and automation as being kind of an inflation fighter. Obviously, we know that demographics and a shrinking workforce are fueling inflation. And industrial automation really is a deflationary force. And robots and automation equipment enable manufacturers lower their marginal unit costs. And so robots, essentially, don't put pressure on labor costs, and that's another way of curbing inflationary pressure. I'm just curious on your perspective on how you see that.
Well, one thing I've learned is how much I don't know about economics, macroeconomics in particular. And so I don't know how inflation works. That's your expertise, Bill. So I have to take your word for it on exactly how that part of it works.
Okay, fair enough. Well, it's just my opinion here that we're sort of approaching one of the best buying opportunities, I think, for robotics really since 2020. And despite a pretty challenging macroeconomic environment and supply chain issues, et cetera, in 2022, believe it or not, it proved to be a record-breaking year for robotics, in terms of orders and backlog. And I think that you've talked about a little bit of the activity you're seeing coming out of warehouse and logistics automation driving a lot of that. And it is interesting that we're either entering, or about to enter, potentially recession where we've got global PMI indices or the PMI index is under 50. And that's happening despite the fact, again, that robotic orders are at record levels. And sort of considering the market trends, I think that probably comes as a surprise to investors.
So I'm just curious if you have any thoughts on what you think investors are missing. And maybe you can also discuss some other areas or bright spots for the market. I know that you have a little bit of knowledge of what's going on in healthcare. It's an area that we think is ripe for disruption, as we go forward, because you have a huge convergence in robotics, AI, and life sciences, that's really starting to bring through breakthrough advances. So just curious on your perspectives here.
Well, okay, great question. And I think where one aspect of the economics of this that's changed is the model of robots as a service. Now this was not... Well, actually it goes back a long way, but it's not that common in standard industrial robot sales. You sell the robot, and then it gets used. And the robot is a very big capital expense and has to be accounted for by the customer. But the new model, and what Ambi is using, is robot as a service, which is where we essentially install the robot, but we own it. And the customer pays on a monthly basis for what the robot does, the service, in our case, sorting packages. What's interesting about that is that now the accounting is moved, because now it's not a capital expense; it's an operational expense. And that makes a huge difference to many companies, because they don't have to put this big capital expense on their books. And they actually see very clearly the benefit. They're paying for it. They can compare it to other costs that they have, and they see that it's actually paying for itself very quickly, so that has helped in adoption. And a number of robotics companies are doing that nowadays. So I think that's one of the ingredients why things are changing.
I think that the costs are coming down. There's a number of other companies that have come out with robots that are making the general cost for the arms themselves, but also the sensors to decrease. So there's a number of good benefits that are coming together. Of course, Moore's law always helps too. We get more compute for less money over time. The other area related to healthcare that you mentioned, I'm also very excited about, because one big change is that there's a number of new competitors in the field, particular of robot-assisted surgery. Now, I want to always clarify that. When you talk about robots in surgery, we're not talking about replacing surgeons. That's not going to happen. I mean, we're nowhere close to that.
But what we're talking about is, how can robots assist surgeons to make them more efficient and more effective? So the difference between an average surgeon and a highly skilled surgeon is tremendous. There's a lot of nuances in how they work. And there aren't that many of the super highly skilled surgeons. So there's this concern of, how can you bring everybody up, the skill level's up? And some of that, one idea, and it's being really explored now, is that these robot systems can learn from the expert surgeons certain procedures, like suturing, and then be able to assist the maybe-average surgeon at performing suturing better. And that's a little bit like driver assist, which we've now seen, right? It's everywhere, just by a Prius and it has driver assist built in. And what that means is it keeps you in lane. If you're about to hit another car, it will slam on the brakes. Those are extremely helpful for avoiding accidents. They're not replacing the driver, but they're making all drivers better. And that's an analogous idea in surgery. And I think we're going to see a huge advance in the next decade.
Yeah, I'm curious on your thoughts on just sort of robotic implementation costs. I mean, historically, they've been high. That's probably impeded some of the progress or some of the penetration rates to sort of accelerate to levels that some would hope. We have seen that iteration costs are coming down, but is it coming down fast enough? Just curious on your perspective on that.
Well, it's interesting. One of the things that we've learned, Bill, is that there's a lot going on behind the scenes. When you are installing robots, you're also the manufacturer of the robots, the systems. You have to get all the components, and we got to source them and bolt them together and get them all tuned and transport it to the location and then installed in that location with the right power source, the right air supplies. There's all these details that have to be worked out. But then it's also ongoing maintenance, because these systems are physical. The suction cups get clogged. Pieces of wiring comes out. This happens. So you have to deal with maintenance, customer service. And you have to be good at that, because if there's delays or if you're sloppy, then the customer gets very frustrated, doesn't want to work with you again.
So these are sort of things that sort of go on behind the scenes. And it's very interesting that those costs traditionally have been... Roboticists don't talk about that, and they talk about their advancing technology. But those are all part of the system to make it really work reliably. The other thing I want to mention is that I think it's really important for roboticists to be careful about overselling their technology. Look, we're all human, and we all want our system to do well. There's a strong inherent bias in anything you do you feel is promising. But at the same time, you've got to report the error modes, the failure modes, as with the success modes. And it's really important to do that, because you share where the advances and where its limits are, the limitations. And that is something I think we need to do a little bit better in the field, because some groups are promising things that I think are a little exaggerated. It could backfire enormously, when customers think this problem is solved, and then they run into problems.
So I think that's another lesson that we take to heart very much at Ambi, which is under-promise and over-deliver. So we really want to build a system and then be able to make people be very happily surprised by how well it works, rather than the other way around.
Well, speaking of over-promising, obviously we know that Elon Musk has pretty ambitious plans to deploy thousands of humanoid robots within their factories and expanding to eventually millions around the world longer term. And he said that robots could be used in homes and making dinner and mowing the yard and taking care of us. And Tesla, obviously, has faced a lot of skepticism in the past. And it's going to continue again now. The question is, when can this happen, a general purpose robot in factories? And the homes obviously needs to come with a justified price. And humanoid robots have been in development now for decades by the likes of Toyota and Hyundai and Boston Dynamics. And like self-driving cars, the robots actually have real trouble, when it comes to unpredictable situations. And they don't have the intelligence to navigate the real world, like they probably need to be.
So there's a lot of outcomes that have to come with consumer robotics. I'm curious on your thoughts on this. And you could almost argue that... I'm not sure what's harder to create the technology for a humanoid or for an autonomous vehicle, but they're both pretty challenging.
Yes. And I think these are areas we want to be a little bit more modest about. I think when we see a robot doing a back flip, then the implication is the robots are very close to human agility, or better than most humans. But it is not true. Those things are very specifically special conditions. The system is trained to do one thing. And then you can take a video, but of course you're not showing the videos where it doesn't work. So it's really important, again, to be very clear about this.
Now, as far as the Elon Musk, I have a huge respect for him. I think he's pulled off really surprising results in engineering in multiple times: clearly with the reusable rockets, being able to stick those landings, very impressive. When he was able to turn Tesla around and be able to produce cars at a cost effectively, also super impressive and really has changed the entire industry. He's also changed the battery industry. And so here's a guy who's very, very skilled at engineering and leading engineering teams. It's a little danger... And this is the old Greek warning. You become very, very skilled and talented and successful, and then there's always the downfall, which is Daedalus flying up too far to the sun or whatever. The danger is that it leads a little bit to overconfidence. And people have talked about that for centuries or millennia.
So I think in his case, when he revealed the Optimus robot a month or so ago, initially, my first reaction was very skeptical. He was announcing that, in a year or two, this is going to revolutionize the economics, that it's going to be used in all factories, and these are going to be available to everyone in their home. And I don't think that's even remotely possible. But what I do think is that he is capable of building and advancing the field of robotics, in the fact that he knows how to build machines, motors, sensors, systems, that are lightweight and reliable and cost effective. So a car maker is in a very good position to design robots. The other aspect is that he has a need for robots in his factories, so I think he's going to quickly find out where they're good. They have to be good at something.
So what I predict is that he will increase consumer confidence in robots. Basically, it's a boost for the field, which is really exciting, because I think people will give the benefit of the doubt. And I think he's going to end up with advances in motors and sensors. And maybe it'll end up being a Tesla industrial robot arm. So it may not be a humanoid, but, in the interim, as that long-term goal stretches out there, I think they'll look for intermediate results. And so something, like a Tesla industrial arm, would be terrific, because we actually do need better robot arms, that are lightweight, fast, safe and reliable. So I'm very excited about his entry into the field, his vote of confidence. I'm a little less excited about his move into social media, but that's another conversation.
Well, just sort of following up on that, maybe you could just help the listeners understand, a little bit more intelligently, how difficult it is to create a consumer robotics system. I mean, essentially you have to model a lot of different outcomes, that we've not been capable of seeing. And that seems to be a limitation that's imposed upon us, and it's going to take a long time. It's going to take a lot of data and a lot of training sets to sort through this. Any comments on that?
Yes. Well, the one thing is that, when you want to work in a very unstructured environment, like a home in particular, the amount of different scenarios that you can encounter is vast, unthinkably large. So you never know. There's going to be a little flap of a carpet that's tilted up. There's all kinds of things that are... These are edge cases. Same is true of driving, by the way. But in a home in particular, you just can't anticipate all the different things that can happen. So what you don't want is this robot that you've bought for your mother, who's 70 or 80, and it suddenly falls over and knocks her on the ground. You don't want that. So in the same way, you don't want a car that's going to swerve off the road and over a cliff. So you have to be very conscious of these edge cases.
And this is a problem for deep learning, because it can work in thousands and thousands of cases, and then there'll be one or two failures. Now those can be fatal, and you have to be very careful. This is, I think, in situations where there are always the possibility of these outliers. And the best example I have for this is look at air transportation, airplanes. We've actually had an automated system, autopilot, for driving airplanes for 30 years. And it works incredibly well, and it's used every day. Well, does that mean we don't have pilots? I don't think so. I don't think anyone's ready to get into a plane that doesn't have a pilot in front. Well, the pilot's job is... What is it? It's to keep an eye on everything, make sure everything's going okay. And every once in a while, there will be a weird situation, like a thunderstorm, and the pilot really gets engaged.
So I think that's really interesting. How do you think about that? And one answer might be something like telerobotics. A number of companies are looking at this, where they have a car that's driving, but when the car gets uncertain, a little stuck, it basically calls a human, who remotely comes in over the wireless network and drives the car, fixes the error. And this can be done for the home as well. So this idea of networked robots, or sometimes called cloud robotics, is very interesting to me. And some people think, "Well, that's never going to work. The time delays are too long." And no, it's not true. The time delays, if you think about when you do Google Maps, basically, your phone is working off the cloud. And so it's constantly getting updates from the cloud, and you don't notice it. It just happens invisibly, and it's very fast.
So this is the technology of cloud computing today. It's far faster and more efficient than anyone maybe think about. But that applies to robotics means that you can have remote computing, remote resources, and put those to use for solving some of these problems. So I think that's going to play a role. I also think there's going to be modifications of places, like freeways, that will have more sensors and [inaudible 00:37:17] internet of things installed that will facilitate these systems. That's going to take time until it's on every corner, but maybe there'll be certain highway sections, let's say, between San Francisco and LA that are very heavily trafficked, and we can put down enough sensors on them to actually have semi trucks be able to navigate up and down those without a driver. But as soon as they get off the freeway, they're going to need a driver to climb in and take it to the destination.
So Ken, given the technological advances, what we're seeing, I'm curious on your perspective from a historical view. When we launched ROBO 10 years ago, we had high conviction that we were in the cusp of ubiquitous automation. And fast forward 10 years, we couldn't be more convicted. And in fact, I don't think that we're in the first inning of the ballgame. I think the players are still in the locker room putting their clothes on, with the exception of industrial manufacturing, which is principally auto, roughly 40% penetrated. Virtually every other segment of our economy has de minimus, or very low, penetration rates. I personally think that the opportunity set, that we have in front of us and automation, is far bigger than I could have imagined. I'm curious if you share that same perspective.
No, I'm really glad you said that, Bill. I think one of the things that... Remember, back in the '20s, when the word robot was first coined in 1920, there were articles about robots taking over all the work. And so what would we do with all our new leisure time? So people have been talking about this for a long time. It doesn't help that television shows and movies often show these humanoid robots doing all these things, and you can't even tell the difference. But that's the difference between fact and fiction. Every time there's a lot of speculation that robots are, "Now, this time, this is when they're going to enter all these new applications."
I think one of the things... So in my mind, when there was this talk, I was worried because I knew that robots take time to evolve. They're not overnight. You have, suddenly, this new capability, and the robots just start working it. It takes time to develop this technology. I think it will come, and I think we're getting it in many different ways, as we've been talking about. And we just have to think about where it's going to happen. And I think in healthcare and being able to deliver material within hospitals, to assist in operating rooms, to assist... I do think it's going to help seniors in homes. I would like that to happen when I'm ready for it, which isn't that far off. But I think it is coming. I think there's a lot of optimism and cause for optimism in the field. But I think you want to think carefully about, "Where is it going? Where's the near term? And what are the more long term applications?"
How and when do you think that we're going to see a more rigid sort of regulatory framework get established in the US and globally, to sort of police the technologies? Obviously, Elon Musk has talked about the need for that to occur years ago. I wonder how big of a limitation this is to a lot of implementation.
That's another good question. I have to say, I've been very, generally in my experience, impressed by how much that the agencies, the care of OSHA and others about safety is actually quite sophisticated. So for Ambi robotics, we have to satisfy many, many regulations, that are very specific about how many feet away can an industrial robot be. How you have a light curtain, so if you break that, and then it has to have a backup system. There's a lot of systems in place across the industry for safety. And systems, whether they're cars or new experimental drugs, are tested very rigorously. So I actually think we have a pretty good regulatory system. I think that we have to be careful. Again, it's about the human users. When we put something out, and we're not clear with the humans, and they think, "Oh, I can take a nap in the backseat of my Tesla now," that's not a good idea. We should probably make that illegal. I think it is illegal.
But being really clear about safety, because I think that the last thing I want to do is have robots, in any way, harm humans. That's the first law of Asimov's law of robotics. So we don't want that. But at the same time, overregulation can really grind progress to a halt. So I'm a little bit mixed on this. I think we need it, but we also want to allow progress to be made.
That's helpful. Well, that kind of concludes my prepared remarks today. I want to thank Ken for his thoughts on the trends in robotics and AI. We at ROBO Global are here to help investors invest across innovation, specifically robotics, healthcare, and artificial intelligence. And we're very excited about where we at. We think that the pause in the markets is giving an opportunity for investors to hit the reset button, particularly as we go into 2023. And we look forward to significant growth in the industry in the years ahead.
Thanks, Bill. Yeah, I think my prediction is we are going to see a roaring 2020s for robots. Let's see what happens.
All right. Thank you, Ken.