AVL's Reimagine Mobility Podcast

Transforming Heavy Industry with Autonomous Technology w/ Bibhrajit Halder

• AVL, North America

Share episode

Join us on the Reimagined Mobility podcast as we dive into the world of autonomous heavy machinery with Bibhrajit Halder from SafeAI. Discover how autonomy is revolutionizing mining and construction, leading to increased safety, productivity, and significant cost savings. Bibhrajit shares his extensive experience in the field, including early projects with Caterpillar and the future of AI and sensor technology in heavy industries. Don't miss this insightful conversation about the advancements and future prospects of autonomous heavy equipment.

 

Bibhrajit Halder is the Founder & CEO of SafeAI, a global autonomous solutions provider for heavy equipment leaders focused on advancing modern industry. Bibhrajit has been involved in the autonomy industry since its early days, beginning with the DARPA Grand Challenge and then going on to work on self-driving projects at Caterpillar, Ford, Apple and Faraday Future before founding SafeAI in 2017. Halder holds a PhD in Mechanical Engineering from Vanderbilt University, and an MS in Mathematics and Mechanical Engineering from Ohio University.

 

SafeAI is a global autonomous solutions provider for heavy equipment leaders focused on advancing modern industry. SafeAI retrofits heavy equipment for autonomous applications in mining and construction to serve a massive ecosystem of industry players with complex needs. The company enables equipment owners to transform existing machines into self-operating robotic assets with aftermarket hardware, advanced, proprietary autonomous technology, and a scalable framework for deployment. SafeAI is headquartered in Santa Clara, CA, and has offices in Perth, Australia, Tokyo, Japan, and Bengaluru, India.

If you would like to be a guest on the show contact: namarketing@avl.com

To the reimagined mobility podcast series. I'm here with Bibhrajit Halder and he is from SafeAI. Bibhrajit you work in a space that everybody's still talking about and it's not EV, it's still autonomous. We're still talking about it. But as opposed to many that we've had the pleasure with welcoming out the show and also that the industry is heavily talking about, which is the way most of this world maybe to cruise it, this world, Lyft and Uber, autonomous vehicle delivery and goods deliveries. You're on the heavy duty side, or at least on the heavy equipment side. And how you help companies through SafeAI your company, make things autonomous. So for our listeners, give us a little bit of a background where you're coming from. And then certainly what is safe AI doing? Absolutely. First of all, Stephan, thank you for having me, you guys. I saw some of your earlier show, you guys doing a phenomenal job. And I tell everybody, like when you see this thing, it looks like all is easy. You were just talking. I know how much what was in the back. So really, really appreciate the work you guys are doing. I think just very quickly, you know what we do, but let me pick up where you started that robotaxis and all one thing is that autonomy been going on for almost 30 years, you know, and I've been involved about 25 of those 30 years. And I've been doing it for over 25 years now. And us, you know, defense industry was using autonomy in 90 like lot of technology, internet GPUs. It's a very similar deal like we are already using. It's that amazing. Can we give it to the non defense use case. And this is what got started. And I was involved when it about this when we got started before any other industry heavy industry actually picked it up before anybody else. So the mining construction picked it up even before your Google. And I'll give you a little bit of history just to set things up. I think Caterpillar and Komatsu's are doing autonomy. Back in early 2000, 2006, 2007, I was part of that. And the reason being in this industry, it's a very constrained environment, high safety concern, remote part of the world, and it's really doing a very repetitive job. Like if you put it all together, it's like a no brainer. Like I want to do autonomy in a more constrained area. I want to do it in area where 24 seven doing the same thing, and it's a very unsafe, you know, so it's a no brainer. That's why this industry started first, just to the statistics out there is about 2000 vehicles fully autonomous, no remote control, human operator human oversight been running for the last 15 years. They move 10 billion tons of material. Not a single fatal accident. So that's kind of the background. So with that background, what we do accept that I was part of the Caterpillar and Ford and Apple. Really we are saying that passenger vehicle as you mentioned Waymo was in leaps of the world. One of the thing they did amazing job. They raised the bar of technology to a different level. You know, we in our world we call autonomy 2.0, the SafeAI. We are taking this latest on the greatest lift. And in a Waymos of the type of technology back into the heavy industry and accelerating their use of autonomy. So let me pause there. We can see. But that's kind of what is our vision. And it was perfect. So and you're alluding to it autonomous or automated as some call it I would call it. It's really autonomous. So for today's purpose let's call it autonomous. It's not necessarily something new. Many people thinks it's been around five years maybe a little bit more. I agree with you. I've worked in manufacturing plants that frankly, autonomous robots were moving things from one line of the assembly to the other side and etc. so not quite to vehicles and certainly not with the sensors we have today. But nevertheless they drove autonomous. They stopped when he jumped in front of them and all of these things. So maybe let's jump in. What what you mentioned, you mentioned that in the heavy duty space or in again, mining, for example, agricultural applications and things that a caterpillar does, a John Deere does, a Komatsu does. You have an ability in a maybe much more controlled environment, clearly, than for me driving from Detroit, let's say, to Chicago. Right. But is it also not that you have a much stronger business case and a much stronger return on investment because that equipment is used, as you mentioned, in some cases, 24 over seven, it's in remote areas, may be difficult for workers to get there, difficult to get companies to send workers there to work. So the autonomous capability needs that are more and more now allowing us to to automize things also benefit probably what you guys are doing, because it is more and more in demand now too, because the business is driving yet and not necessarily out of a convenience type like so maybe it's yeah, a little bit of thoughts on those things. Yeah. And I think the comments you started making that you have been involved in, you seen autonomy long time and it shows in your question, right, that you nail the two in a two main point on this thing. First thing is that it is lot more easier, not more repetitive. One of the biggest thing is that this is a private road, right? Think about a big mine, big construction site. So it's a private. The difference between private versus you driving from Michigan to somewhere is that we get to make the rule. What I mean by that, back in the days in 2012, 2011, 2012, we could not solve the problem of intersection management. We like okay, wait, if we are mineplex, how do we manage that? No, we didn't have enough juice in the onboard compute, so I cannot write that algorithm there. So we said, you know what, let's make it easy. When there is autonomous cars in intersection, everybody don't have a right. Autonomy has a right. So I cannot do that in Michigan. I can't do that in Detroit. Like I can't say my autonomous car will always have the right people. Like, no, this is a public road. You have to. So I think that gives you an immense advantage as a technology provider, right, because you can control the environment. Second thing, what you are saying is absolutely right. And the way I say that autonomy is not about who needs it or who can use it, it's about who can pay for it. It's a it's a high quality, high cost kind of technology. Each. And think of a mining industry, they do run 24 seven. They pay their operators half $1 million per year, per truck. And you compare that with the Uber about 60 K 40, 50 K whatever the number may be per year per truck. So now you are competing against half a million versus let's say 50 K who can pay for autonomy? let's say 50 K who can pay for autonomy? Obviously the half a million. I think this is why why you saw heavy industry picked it up. So two point you mentioned is absolutely the two critical point interesting. And so the touching upon a point that's been a big topic, I think generally speaking, as autonomous has still to shift from the 20 1819 timeframe when everybody in the passenger vehicle space said by 20 2021, blade is 22, you'll drive in an autonomous vehicle, right? And that kind of that time frame came and it went and nothing really happened. And everybody kind of scaled back. And now is focused on on ADAS, level two and three. Right. Not necessarily four and five. And it's sort of I saw a shift maybe 3 to 5 years ago to the heavy duty truck side. Again return on investment much quicker maybe but also clearly with the mindset we need to do something in heavy duty trucking on road, not off road on road. Because we have a tremendous, driver shortage coming up or is already here. Do you see the same thing in the space you're in, in mining that you just mentioned? It's a cost, right? It's a big cost. Your pain in operator versus investing in, in a autonomous system that then allows you to do that not only 24 seven, but for the next 2 or 3 years straight. Is is there a labor shortage, so to speak, in the areas you work with that gives the gives autonomous technology in your sector a further boost in a further importance as it relates to investment money flowing in, technology companies and operators focus on and etc.. What do you see there? Yeah, absolutely. I think they'll give you the labor shortage kind of very acutely. I mean, think about in our world, we work in a remote parts of the world, a construction site, let's say you're building a new area, you're going to new my these are never in the middle of downtown, right? It's a far away anyway. It's hard to get people there. Safety is a concern, a huge concern actually. Fatal accident happen significantly in this. Much higher than any other industry. And the third part is that the productivity. One of the thing is that this autonomy actually provide 20% productive productivity. You know, if you look at from the mining and construction point of view, let's say they're doing a project like $100 million, five year, they can get the project done for $75 million in four years. So it's a fundamental shift. So they're both I mean, when we go in front of a mining and construction company, they're not asking why autonomy. They're like, how much are we going to charge? Do we have a support? If my fires break down, who is going to do my part? Supply? They are asking like a real question. So you would not have any resistance. Any mining company taking autonomy if you can provide. So that's what we are I mean really in our industry we are the technology provider that if we had all it ready, would have go. And that's really our vision to go as fast as we can, providing the solution to the industry. Even at a general citizen. Forget about mining, construction the way you want to think about, you know, your city build stuff, right? Airport, bridges, roads and all. And you see there take five year, you rather get it done it four year, right? And less money and where do you think the money come from? Your taxpayer dollars. So it is actually have a huge impact on the society. And it's being on a secured on a scale. Back on the passenger side, you mentioned robotaxis and the trucking. The scale back have been very, very interesting analysis is that investment dollars actually do to scale back. What scale back is that that investment dollars was pouring into 60 company. So there is a lot of noise a lot of buzz going on. So they they're all kind of getting distributed. That 60 becomes 6 or 5. That said there's a bus went down but the dollar did not go down. Right. If I'm spending 29, a $10 billion into this 60 company, I'm still spending $10 billion. But now on spending on those five company. And that's actually good for the industry because it's consolidating. And you can see the winners coming up. Right. Waymo is still running really well. You got the Zeus good going well. So you will see lesser and lesser played. Because autonomy is never a game of 60 players. It is eventually going to be 2 to 3. It's like your iPhone versus Android and whatnot, you know. So I think you will see the consolidation and that will be massively beneficial for the industry. We don't need seven different companies riding the same algorithm, paying the engineer same money and just making the engineer life better. But that doesn't improve the society. So I think it is actually scale back on a number of companies, which is actually a good thing. Now, I agree with you that, you know, having 60 solution providers makes it very difficult for customers, right? Let's say equipment OEMs or passenger vehicle OEMs to figure out which one is really the right one. Right. And I think dilutes somewhat from maybe pushing forward because it's difficult to figure out who's on first, who's on second, and who does what. I mean, I've mentioned this many times on these podcasts here that over, you know, with the last 2 or 3 years. And I was at CES So I always come back and I'm amazed again of seeing even more lidar providers, yet very few applications of lidar in mass are actually out there. And you wonder, how are all these companies a distinguished themselves? But be survive in a market where again, you have so many offerings, yet not too many OEMs really have production programs going on with LiDAR. But what? Let's shift for a second to the technology. Let's talk a little bit about as we reimagine mobility here together, take take our listeners and viewers through what you see happen over the last five years or so from, let's say, 2020, 2019 to today, how much and how fast has technology advanced sensors. You mentioned compute platforms. We have the ability for simulation. And then let's flip it and look for a word and send it backwards. What we're really interested in as we again reimagine mobility. What do you see happening over the next five years. So a little bit give us, what happened over the last five years. How far have we come? And then from your perspective and your company's perspective, where are we going next over the next five years? Yeah, that's a very good question. And maybe I'll take a little bit of more than five year, ten year just to set the stage out when it got started right back in the days 2012, when we had a very limited sensor and we have a very limited compute platform, and it was a very hand-coded like we all wrote, like if this happened, what do this right? And we did, we did good job. But that's all we could do. That's why I don't say it's a bad thing, right? Because I wrote those code. So I cannot say bad thing about it. But that's all we had like one shift happening in 2015. 2016 is the date I came along. The GPU came along. All of a sudden we have a lot more choice, if you will, the processing power. So algorithm improves significantly, right? Third shift happening now, which I love, is that now the cost of sensors going down right when I was using back in 2010, LiDAR was still 40,000 and $50,000. 2018 2017 is still $40,000. Now I'm buying lighter for $4,000. Much better. LiDAR than I used for those $40,000. Right? So one of the things happened in the technology, the cost is just going down significantly, which is very good for the consumer, very good for the scale up. Second thing happening is that the AI really now getting to the point where it was fun, it was good. It is working. But the way we use AI, we have saying that let's use AI, but let's put a boundary around AI in the algorithm with a deterministic algorithm. You know, this is where our name also can say AI. It's not just us. The whole industry was doing that way. Yes, I'm going to use AI for XYZ, but I'm also going to put a boundary around that with a, you know, like a deterministic algorithm. Now he AI is really shiny. So I think what you will see that the cost will go down significantly, which are already going down. And your algorithm take it another notch up. And that will really give benefit to in our world, in the heavy industry, what we will see. They used to pay about $1 million for the hardware kit. just think about $1 million for a harware kit because their truck cost $5 million. That million dollar is going to go down less than 100 K. And imagine what it does to the consumer, right? The companies are buying say that instead of paying million now I'm going to pay in less than 100 K. The algorithm was good for very simple use case. Now we can say I can do a lot more complex use case for you. So not only mining I'm going to do your construction. I'm going to do your, you know, cement quarries. I can do all those complex because my algorithm went up. So now they can also really expand because some of those companies are not just a mining company. They are such a big data mining construction. They're like, did it like 100 plus billion dollar company. So in our industry I think at SafeAI our vision and what we see is going right now about 2 to 5% fleet in autonomous in this world. We want to take it to 40 to 50%. That's kind of where our vision is. But in the passenger in all that side, you will see that, you know, Waymo is doing a great job. The slowly doing the job, right, because of the cost and maturity of technology, is increasing. So you're bringing up a very interesting point. You mentioned data rate. I mean, I've made a statement probably five, maybe eight years ago now when Google first came out with their autonomous vehicle, I made the statement. I said, I'm not sure they're interested in being coming in autonomous system provider, but they're doing it because they want the data, because Google, at the end of the day, is a data company. And how they can take that data and use it as a sensor is probably the most powerful sensor in the world, because it's a vehicle that drives through all sorts of conditions and collects who knows how much different types of data and how you can use that to feed it back into your main business, which is search results and and and revenue derived from providing people with information in your space in the heavy truck space. I understand they got to move dirt. They gotta plow a field. They got to move concrete from one place to the other. How much of them are actually talking about what are we doing with what should we do with this data that we're collecting as part of the autonomous system that we have installed to allow this equipment to work autonomously? Are people talking about that, and if so, do you have a few examples of company that are actually doing it? And what is. Yeah. And you're right. They are talking about it. I think that's for sure. What they're not doing well is that they're not actually translating the talk into an action. And the reason being, they haven't seen that use of the data actually put into use. And, you know, I think our industries still talk about when they talk about data, they still talk about diagnostic prognostics and vehicle upkeep and all that. And I tell them, those are those are babies job because you haven't seen data year one of the thing, And I'll give you an example of a construction project. Let's say you have done $1 billion project of airport rebuilding. Now Singapore wants to rebuild or expand their airport. They put out $1 billion kind of RFI. If you wait and all the GC ... general contractor bidding for it, one of the things they do, they have to guesstimate, like what would be the cost of that? Because this is the bidding process today. They do the bidding process, all because there is a five guy into this 100,000 company, you know, knows how to do this airport bidding, right. And then them, then they update that. Think about you already have done the airport project before. If you have the all the data yet now you can guesstimate so much better right. And this is where they lose or win money. Because if you guesstimate you're bidding wrong for $1 billion project, your profit margin can impact. I think this is it. I say this now you understand your diagnostic prognostic look small and this is where they're going to start using like I have used this project, I know exactly what I ran because my machine are smarter. They're not lying anymore. If they are saying I'm going there they actually went there. It's not a noisy sensor. I can absolutely use the data to predict my next project. And also it goes to supply chain is to go to procurement. It goes to people management literally tools for to their ERP system. And the beauty of this world is that, you know, that is that they haven't used the data before, but because they haven't used the data before, the advantage they can get, there's a lot to squeeze. There's a 30 to 40% efficiency. They can get that. They barely squeeze any, you know, any juice out of that like compared to other industry. So I think the heavy industry will see a massive, massive benefit for the data. Interesting. Yeah I would agree I would agree. So Bibhrajit , let let let's take it personal to you. I mean, you you've mentioned before you've been involved with DARPA that's already many years ago. You've worked with Ford with with many place player sorry with many players in the autonomous space and you're clearly passionate about it. What are you mostly you not necessarily SafeAI or your people, but you specifically, what are you mostly excited about seeing happening over the next five years? You've talked AI, so maybe it's, you know, even more deep learning, right? Large language models, the combination with the hardware that I still feel is somewhat a little bit gets a little bit push to decide it's all about AI into software. But the thing don't run by itself. It needs hardware. So that's still important, right? I'm excited to see what comes out there. But what are you most excited about in the space that we're talking today? Personal. Yeah. And I think just do it. And I and my first autonomous drivers in 2008, in a cat 793, you know, this is a massive big cry. That was my first autonomous truck. I probably, you know, being in autonomous, all of the company you think of. Right. And I used Tesla for 2060 before it even came out because I had a test vehicle for them. Right. I think the experience of actually using autonomy on a day to day life, it will fundamentally change. So that's kind of the really exciting second thing, if you think about transportation as a pillar of society, you know, you know, think about food, transportation, medical. These are pillar of society. And we're going to stand on that. Right. We are fundamentally change the transportation with autonomy. So I think going back to what your whole this podcast about reimagining, really we will change the fundamental like 5 to 10 year how we do transportation. Right. And you think about if we change one pillar of the society such a profound way, it will have a huge impact. And I don't know what that will be, but I that's I'm really excited about where that society will go because we are touching one of the key pillars of the society that humanity stand on. So that's really exciting. I think technology and all thing will happen, but we really will leave and do things very differently going forward once this technology is more profoundly actually impacting that that linear path. Well, good. What good will I get it. So so to your question about the right one, and I think it's a great I wasn't thinking Tom, but maybe not as much anymore. On when EVs first came out. Still a big topic as it relates to cell. When we talk propulsion systems, I think on the left side it's like the challenge as well, which is now getting to a point where people are comfortable with it and it's all about this consumer acceptance, right? Weighted ahead. I don't want to talk passenger virtue. I don't want to talk heavy duty trucks. I want to talk to your siblings right now. So this heavy equipment and we've talked about these companies that maybe get investments in it because see, the benefits. How many of those customers again or the companies just competing and how much acceptance social media. Right. How much should we license or uncertainty or skepticism. You still see there. You know, what about maybe the operators of the maintenance people that still are somewhat involved or the people that work besides this and those autonomous equipment or vehicles? But how much afraid or embracing are they to these technologies share a little bit about, again, the consumer acceptance point here and yeah, and I think there are three major kind of the stakeholder, if you will. One is your operator, the people who are actually running this vehicle doing their job to get their company going. Second is the people who is actually doing the job on the ground. A third is also the government and the government body of that country. Right. These are the three player has to kind of get comfortable with it. I think the operator level people, like people who are running the company, who are running the site, doing they have been sold since 2010. There is no CEO or CTO or CEO of a mining company. Say, I do not want autonomy like that. Was that debate was done in 2010, 2011. If you look at BHP, Rio Tinto really report you can go all the way back to 20 1516. They have been consistently saying, I'm getting 10%, 20% product improvement. My safety number is zero, right? So they are absolutely sold. I think people at the beginning was skeptic and I can give you a lot more example. I have seen first hand when we are doing the, you know, integration, but now people are a lot more actually accepting it. And treating is right. There is a shortage of labor. But second thing is that as the aging population go, the new generation doesn't want to even do that. And I was in a mine in Australia and I was driving with an operator. She was about early 20s. Right. And she's in Australia. Australia paid an operator. Very good job. That's why she's doing that. She like but I'm not going to do that more than 3 or 4 years. You know, I'm going to do this thing. I'm going to make my money and I have my other job, other career, and I'm going to do that. This is not I'm going to continue. This is a very good for me to learn this industrial Australian mining is big. So I think next generation doesn't even want to do this thing right. I think so that there is a huge acceptance coming. Maybe there is a the people who are now, but there's the older age like average age of an operator is over 45 years. I did it changes. And the third is the government in the it will be a mix. Australia, Canada, US, Latin America government is super supportive. Now we had we are in Japan and some of the other countries governments are really supported. They see the benefit. So I think in our industry they accept in these three pillars, they're working really closely and operators are really sold. People are getting much more Walmart and the governments actually supporting. So it's a much more positive environment than your passenger side because this is like passenger is a lot different. Right? Just because the question comes in, are you going to put your kids in that car? Right. That's a very personal that touches your home. Right? It's not a question of business decision anymore. Now you're talking you're making a decision with your heart. Like, am I going to really put my two kids in that car? Right. That's a hard decision. That's not a decision. Yeah, and very interesting. Right. And final question to wrap it up, nothing to do with your industry. But again, with you, as we talk mobility and as we reimagine mobility, even for you, what is going to be the next car you're going to buy and why? Okay, so I'm biased. I'm in Silicon Valley. So my current one is Tesla. So I think my next one is probably going to be Cybertruck. They have a two year wait time, which is good. I told my wife she's like I wish there was time five year. I don't want you to buy any more. Got. But I am very the Tesla I've been driving since 2016. I love it so probably Cybertruck if she allows me. If she allows you all right. While you are. Everybody has their boss. Yeah, we already have it, but I know how you feel. I'm in the same boat. Very good. Thank you so much for your time and really shining a light, I think on a on a an a on a sector of the autonomous space, that very few people think about. But as you mentioned, amazing. Maybe to those same people that don't think about it is well, that's the space autonomous systems have been used probably for the longest, very successfully, as you pointed out. So thank you so much for your inside here. And, thank you for contributing.