This interview with Mel Torrie is all about autonomous vehicles. Mel is the founder and CEO of Autonomous Solutions. Mel started Autonomous Solutions in 2000, bootstrapping the company into a world leading autonomous vehicle solutions company. They help mining, agriculture, automotive, government, and manufacturing industries with remote control and fully automated vehicle solutions. Mel was ahead of his time. They’ve automated 75 different vehicle types, which is pretty amazing.
I asked Mel to be on the show to hear about the history of Autonomous Solutions, how the tech has developed since 2000 and where he sees it going. I’m also curious what he does to feed his basketball fanaticism.
Some other things we talk about:
– What was the first autonomous project you worked on?
– How is your safety record? It’s impressive, super impressive.
– How much money has been contributed by vertical development partners? It’s a huge amount.
– How does your tech work?
– What problem will self driving car companies encounter?
– What problems are you especially excited to tackle?
Dave Kruse: Hey everyone. Welcome to another episode of Flyover Labs and today we get to have Mel Torrie with us. And Mel is the Founder and CEO of Autonomous Solutions. And Mel started Autonomous Solutions is 2000, bootstrapping the company into a world leading autonomous vehicle solutions company. So that means they help mining, agriculture, automotive, government and manufacturing industries with remote control and fully automated solutions. So I guess Mel is quite ahead of his time. So they have automated about 75 different vehicle types, which is pretty amazing. So I asked Mel to be on the show to hear about the history of Autonomous Solutions and how the tech has developed since 2000 and where he sees it going and if we have time at the end, I’m also curious what he does to feed his basketball fanaticism. So Mel, thanks for joining us today.
Mel Torrie: Great to be here.
Dave Kruse: So, yeah, like I said you were definitely – I mean everyone knows about autonomous vehicles now, but didn’t back in 2000, but before we get into that, can you maybe tell us a little about your background and how you came to start Autonomous Solutions.
Mel Torrie: Sure. I was raised on a farm up in Alberta, Canada in the middle of nowhere and drove in a circle 16 hours a day on the farm and definitely some of that kind of mind numbing work motivates you to get an education and then just happen to work out that even getting that education we were able to discover some ways to automotive that driving in a circle 18 hours a day, so kind of a fun journey there. And so went to Utah State University down in Utah and John Deere saw a paper I wrote in my graduate program and asked us to start doing autonomous agriculture for them and so we did two years of work for them, and then they asked us to start a company to partner with and we spun out in 2000 to start doing agriculture robotics and then diversified into mining and cleaning and security and these other markets that we are in now, so kind of a general story.
Dave Kruse: Interesting. So when you are – so how did you get into autonomous vehicles? Was it in grads, was it in school and why – what prompted your interest? Was it being on a farm and seeing the monotony of driving around a vehicle.
Mel Torrie: Yeah, it’s a good question. I think it was always hard to choose between computers, electrical and mechanical engineering and one day I saw a wheelchair driving by itself down a hallway in my undergrad program and started to chase it and find out who the professor was and started begging him for a job and finally got in sorting bolts into the lab and then they paid for my graduate work, and it just seemed like a great combination of all of the technologies, the software side, the mechanical side, the electrical side and so I just gravitated towards it, got excited about it, and then it was really that coincidence of the hallway that really got me excited about it.
Dave Kruse: And when you started back in 2000, who else was doing autonomous vehicles? Was there any other folks?
Mel Torrie: When we spun out or when I started in robotics?
Dave Kruse: Oh! That’s a good question. I guess when you – well, either one, but when you – I think it’s when you spun out.
Mel Torrie: Yeah, I guess it was still pretty young. I think CAD had started back when I got into robotics and then the early to mid 90s and then they had started working with some construction equipments, started playing with some mining trucks. That was kind of the state of the art at the time and we were doing it more under department of energy funding. And that was a government initiative to look at handling nuclear waste and so that was really the – the state of the art of that time was hey, we need to address this hazardous material handling and so the government was willing to spend money at that point. So that was really the start of the funding at the lab that I was a part of. And then there was some early workings at the Carnegie Mellon University and in their rack within the construction mining site with CAD at that time and that was the kind of the state of the art at that point. They were starting to patent some things and do some private trials.
Dave Kruse: Okay, but not a lot. And before we get too far, can you maybe just give an overview on Autonomous Solutions, like the number of employees and you know maybe a description, unless I did enough justice, but yeah, it might be better to hear from you, before we get too far.
Mel Torrie: Yeah. We are over a 100 people. We are actually a group of companies for liability isolation and investment flexibility. So I have an ASI Mining company, an ASI Cleaning, an ASI Security an ASI AGG, an ASI Automotive and so we have bootstrapped from the beginning. We haven’t had any investors’ to-date and we’ve just been funded through convincing monster OEMs like the John Deeres of the world and monster companies like big mining companies to fund us to develop technology for them. And so I think we have raised over $85 million in that kind of money where there is no equity, but it’s strategic and very blessed that way to have been able to find people who were interested enough in betting on us in the middle of Utah to bring these kinds of products to the market. So very fortunate in that way and we basically run Agile/Scrum, so we have teams in each of those markets with business leaders that are growing each of those markets for their full potential and yeah, so we got about 100 acre proving ground up in Utah and basically test mining trucks and farm tractors and security robots and floor scrubbing robots at our facilities and in the surrounding area.
Dave Kruse: You must have one of the more cooler office/locations there are, I would think.
Mel Torrie: Yeah, it’s pretty fun, yeah. Yeah the command center was – we have ongoing durability testing for functional safety and product evaluation, ongoing all the time. So people should show up and there is always robot vehicles out running, whether it’s on the outside test tracks or out in the farm field and there is a command center that we can look out through the big windows to see the equipment running and look at the software and change the machines and things like that on the fly.
Dave Kruse: So I love the story that you did bootstrap it and that you didn’t take outside the equity essentially or cash for equity, which is pretty amazing and that’s really hard to do. And you know I’m curious if how – if you can’t disclose this, that’s okay, but how you structure like those joint ventures. You know let’s say you partner with an agricultural firm, would they get like the exclusive on the technology or how would that work?
Mel Torrie: Reveal my secrets? Yeah pretty – no problem at all. Pretty standard agreements where we were able to get to the point where people wanted the technology enough and there aren’t many providers out there that we could get pretty good deals. So we provide them exclusive rights to that vertical and so let’s say we are negotiating for a lawn mower contract with an OEM that does lawn mowing. We would give them exclusive rights to any IP developed with any of the other verticals for that market and we would jointly own that IP. So they would own it and we would own it and have the rights to use it any other markets. So that’s what some of these companies really like is that there is millions of IPs coming in every year from other markets with IP they can actually leverage. So mowing for example, we’ve got indoor cleaning going where they don’t have GPS. So they are doing just slam based navigation with lasers, no infrastructure. Now that mowing guy gets exclusive rights to use that outdoors. So if he is near a building, under trees, in a golf course, near a house, that navigation that was paid for by a cleaning OEM, now he has exclusive rights to that. So he doesn’t have to pay the full bills we are getting. We have many of these verticals that are all contributing multiples millions a year each into an IP portfolio that they all get exclusive rights for their vertical. So that’s one of the big pros of how we’ve structured it and why they liked that kind of multiple they get on their money.
Dave Kruse: That’s smart. So with the mowing example, would – like I have a couple of questions. One is, you get paid for development. Do you also get a royalty on that? And then like can they sell it, do they sell a technology to other mowing companies or do they keep it to themselves?
Mel Torrie: No. So there is definitely some guidelines around the transfer. We take that seriously, but yes, we get licenses for every unit they sell and they can use it throughout their market. If they want to sell those rights to someone else, that gets a little scary as you don’t want to violate any of your other agreements. And so if they sold it to a big OEM that had lawn mowing as well as mining drills, that’s unlikely, but they wouldn’t be allowed to do that, because then they would violate our exclusive agreement in mining. So we just have to be careful, especially if it’s a source code escrow kind of thing that we are not violating any other partner agreements.
Dave Kruse: Oh! Interesting. Well, that’s a great way you structured it, yeah. So I was curious, what was the – back in 2000, well it sounds like you were working with John Deere, but what was the first project you worked on around automating?
Mel Torrie: There were two. One that we had started in my basement, which was the robot called Chaos, which is a walking tractor vehicle and we convinced the military to fund those through some SPIR grants and then in parallel we were doing the John Deere Orchard Vineyard Spring Tractor. And so getting farmer out of the chemicals driver was the motivation there and there was definitely customer demand and so those were really the two that began. It was more of a military rescue type ground robot with an arm on it and high mobility and then this farm tractor.
Dave Kruse: Interesting and then so with the orchard tractor, was that completely autonomous or was it guided by…
Mel Torrie: Yeah, completely autonomous.
Dave Kruse: Really? So it would go up and down the rows and spray and then go down the next row and spray.
Mel Torrie: Yeah, yeah. So we have done some of that with the Department of Energy at the University where we had multiple vehicles out coordinating spring type applications for neutralizing a chemical spill or something for the Department of Energy and so when we transitioned out and brought those kind of people, John Deere said we had to ditch the university like computers and operating systems and coding approach and move to more professional products type processes and so we brought those kind of people who knew how to do multi vehicle planning and coordination and so if one of them got a flat tire, the other one will take over and handle that kind of an orchard emission and so that was definitely the idea there.
Dave Kruse: And how, so I am curious how the technology has changed. Like can you describe the technology in order to implement that orchard tractor, than compared to what you – how you do it now if it’s different?
Mel Torrie: Yeah, it’s pretty similar. It’s kind of embarrassing, I’m not sure if it’s embarrassing or brag to say how long it’s taken us to truly get thousands of units out in the field. Because very similar, the cost points have come down on the LiDARs, but we were using LiDAR 19, 18 years ago and we were using radars, and we were using computers, definitely bigger computers and more of them, but very similar sensor technologies just lower cost, faster, high fidelity. You’ve got millimeters instead of centimeters kind of think. But general evolutionary changes to the technology and the drivers and cost, but not very different as far as the algorithms; if there is something in front of you, stop, if you are in an orchard or if you are in a field, you don’t want to be chasing a cow and messing up your crops, so some pretty basic approaches to the intelligence of the vehicle.
Dave Kruse: And so do you ever use the GPS too as part of that?
Mel Torrie: Yeah, yeah, they were definitely more expensive. You were paying 50 grand for an RTK solution where – I mean you were getting nice four inch accuracy kind of things with your own base station, yeah, but it was just incredibly expensive. So now you are, you are in almost in order of magnitude, 50 grand, now you are at 5 grand kind of price point. So that definitely starts to make it more practical for industry wide sales and net income.
Dave Kruse: And what’s one of the favorite projects you worked on as far as autonomous vehicle?
Mel Torrie: Well, I think that chaos robot was definitely my favorite. That’s a sad story because there was never a single market that had enough traction in it for you to get the cost point where you could sell a lot of them and so it’s our most requested robot and yet we could never invest the money to get the cost down. So that’s really my sad story. But I think for excitement and fun, I think the golf course mowing we did with John Deere was great back in 2001, 2002. Definitely John’s was worth a low cost positioning when you are looking at golf course mowers who get free golf and work for free or minimum wage, so they can get free golf passes, very hard to compete with them. But that was where you are really getting for the fidelity of an inch or two accuracy, making the really cool stripes in the fields and we really got people excited about the technologies and the potentials. So that was definitely a fun era.
Dave Kruse: And you know with – let’s just say with that project, can you give us an idea of how long does it take from beginning to end to finish a project like that and if you disclose cost, that would be great, but I understand if you don’t.
Mel Torrie: Yeah I think, let’s see – I think timeframes now, you’re in a – it depends on the vehicle and the platform and the level of drive by wire automation. But typically you are in a six to 18 month kind of window for getting a product out there. So the new tractors you get are CAN bus controlled, ISOBUS Class 3 kind of protocols where you could plug into that and have it automated in a half an hour and because everything is commendable through that message set versus I’m going to take an off the shelf forklift and turn it into a robot or make a vehicle from scratch, and those definitely have a much longer development period. So it depends on the platform your chose and the level of sophistication and electronics.
Dave Kruse: So if somebody came with you with a new vehicle, a new industry and I can’t think of one right now off the top of my head, would that be a lot of – I mean it sounds like there will be a fair amount of overlap, and I’m sure there has always been use cases, but are the projects a little, a lot faster now too because you already have these existing algorithms and the code base.
Mel Torrie: Yeah, I think, we definitely try to make the platform generic, meaning the building blocks we have for the software or the plant control, the vehicle software intelligent and the vehicle brain computer hardware that is quickly adapted to whatever new application you are looking at. So we use the same computer brains, whether it’s on a bulldozer, mining trucks to a lawn mower to a floor scrubber, so that’s very quick. I think there are definitely those corner cases, like hey, in a farm you want to stay away from obstacles and in a cleaning – a Wal-Mart you want to get as close to the obstacle as you can and get it clean. So some scenarios like that that are a little bit different, and then the other difference is the safety guidelines. So if you are ISO 2662 in an automotive space there is some other ISO standard that we are working on for mining of for Ag and they will have a little bit different tweak in functional safety and the development guidelines and perhaps the color of the beacons or the process – the emergency stop protocols and interlock. So there is definitely some corner cases on the application and then some safety parameters to be complying with the standards.
Dave Kruse: And I was curious about the safely, some of the equipment you are working on is not small and all of its pretty much dangerous. I mean how do you ensure – and of course we read about like the tests are crashing and that’s going to happen. How do you handle safety and have you had any accidents over the years?
Mel Torrie: We haven’t had any accidents. I mean when you are testing something new in your own proving ground, there are definitely hey, we are going to test this new algorithm and it didn’t steer right and stay on the road and then you pull it out of the ditch and put it back on the road. But as far as release to customers or anybody getting hurt at our facility or outside, there hasn’t been any issues. So the, I guess the approach, that does really come from the aerospace and the automotive where they’ve done exhaustive analysis on best practices for developing radar and systems that keep people out of harm’s way and so they are very standard approaches and most ISO standards come from those two industries that then are adapted for mining and then for Ag and then for cleaning. So they have done a very good job at the analysis of the risks, analyzing the likelihood of that risk, the seriousness of that risk and putting the miracle analysis around that, so as you design a system. You are addressing it with design and with standard operating procedures to make sure that no one will get hurt. So it’s definitely hard to predict what humans are going to do, but you try to follow these new miracle analysis so that you can show you followed best practices, you’ve done everything that the industry has indicated as a safe approach to design and just hope that people are wise and there is an example of nice score where the little kid tried to give him a hug and the security robot, that is not a scenario you typically think about.
Dave Kruse: Right.
Mel Torrie: And so we have a security robot that we are releasing with Sharp out of Japan and so that is one of the safety cases that we have to look at is, if that kid really wants to have an intimate hug with that robot or some kind of interaction, how do you address that safely. Those are imperial scenarios to address in the right way and we do our best our best following the best practices and hope for the best.
Dave Kruse: And as far as running into something, I know you’ve talked about redundant systems. So would that mean – I know you have LiDAR, but would you also have another sensing tech behind that in case the LiDAR didn’t pick up something or…
Mel Torrie: Yeah, it’s different for each system, but in the security space we have more of a 3D camera. We have a bumper and then we have a LiDAR. I think those are the three levels on that vehicle. Then for other applications like mining, we’ve got LiDAR and then we’ve got radar and sometimes straight out depending on the speeds and the dust levels in the environment. So there is typically some different type of technology that is used to provide another level of safety of some sort.
Dave Kruse: And can you give a – well, this is going to be hard, but a brief description of how the tech works. Like whether is for the security vehicle, another one. I mean you have all these sensors that you bring this data and you analyze it and then the vehicle makes a decision and can you kind of – is it possible to walk us through that without spending maybe an hour on it.
Mel Torrie: The basic intelligence of navigating in its world and accomplishing its objectives.
Dave Kruse: Yes, yeah.
Mel Torrie: Yeah, so it’s pretty similar for all of them. But a mining truck, it is given its high level – we call it artificial intelligence engine, but as the new truck is powered up and they want to move it into an AI assigner for the certain area of a mine, then that Artificial Intelligent agent takes over that and has control over that vehicle and so it might have four or five already or 10 or whatever. It will start to do the analysis of, ‘I need you to load from here and I want you to dump over here,’ and it will optimize the traffic path and then send that path down to the vehicle. Now that vehicle starts to navigate from that shovel to that dump site and as it encounters difference like there is a truck stopped in the middle of the road blocking it or that intersection is blocked, I need to turn around and dynamically avoid the kangaroo, those kinds of real time decisions are happening onboard the vehicle. The high level of control said that maybe you just need to get the stuff from here to over here and then as that vehicle navigates that, its local intelligence onboard the vehicle and its sensors provide the information to navigate those kind of real time surprises to still accomplish its mission. And so it will go and dump the load of gold and then it will go back to the shovel, again planning the optimal path and the shovel, it will back up to the shovel. That guy will then load it manually and then he will push a button saying your loaded, go dump again and then it will go through the same process, so it’s simply navigating that.
Dave Kruse: Interesting. That’s a good description, and that was off the cuff, that was pretty good.
Mel Torrie: You know so there’s levels of intelligence and there are people to describe it, but there’s intelligence at the high level and then there is intelligence on the vehicle and your also doing this traffic management we call choreography, which is you want to maximize the flow of gold out of that mine and so if you have service trucks or unloaded trucks, you want the loaded trucks to have the priority through the intersections and so there is this speed, this dynamic speed control that is changing the vehicle speeds in the area to assure that the way the trucks always have the priority and the unloaded truck will slow down, so it doesn’t cause a block at an intersection and things like that. So there is multiple levels of intelligent that are all pushing towards safety and optimizing of the operation which maybe gold or maybe farming, getting the wheat out of the field whatever, but optimizing those with the best algorithms that you can come up with.
Dave Kruse: Interesting, all right. Well, we are getting near the end of the interview. I’ve got two or three more questions and one is, I mean it’s kind of about the future. I was going to ask you about what’s your vision for the future of Autonomous Vehicles, but that’s kind of a – you could also answer if you can, what projects were you looking for to work on really. What’s interesting you right now and yeah…
Mel Torrie: I think there is a couple on my radar right now and mowing is one that we’ve got three or four mowing manufactures talking to us. Some are small at niche mowing and some are very large consumer side and so landing a deal there. I think the people mover is an interesting one that is getting some investment right now where you’ve got these driverless busses moving around Disney Land, that kind of thing. So we’ve gotten some inquiries there. And then the third one would be the food delivery in town. I think we’ve got some inquiries – I don’t know how serious they are and how much money they have, but I need to deliver pizza to this address. Some of the challenges of navigating city streets with traffic lights, that kind of thing brings us some unique challenges that will leverage our indoor navigation, that will leverage the obstacle protection and avoidance and all those things. But now I’ve got traffic lights and cross walks where people are making right turns on reds, even though I have a walking sign. That kind of thing looks like a fun challenge. I want to see if we get one of those deals.
Dave Kruse: Interesting, okay. And what – so of course self driving cars are all over the news. What advice would you have for those companies or what problems do you think they are going to have that maybe they are not foreseeing, because you probably have more experience than almost anybody on the plant, with autonomous vehicles.
Mel Torrie: Yeah, I think the challenge is definitely those corner cases, those exceptions that – and so I think the [Cross Talk] yeah, exactly, that funky thing that you have never seen, even in your 2 million miles you could go out on the street and set up a scenario that’s never seen. How many millions of miles do you get to before you don’t hurt someone in those exceptions, because I can model a lot of these environments that I’m already in for most of the things that will happen. But now I start driving down a city street with kids and people playing in the yards and those kinds of things. There is just a whole another level of complexity beyond a farm or mine or Wall Mart and those kinds of environment. So I think that is the challenge there. It’s hard to wrap functional safety and lien time between failures, when you have the randomness of an urban environment in communities where you have people on bicycles and all of those. So I think that’s definitely a hard nut to crack. It’s going to take a lot of time and I’m going to bite off the low hanging fruit first. I wish them luck and I’m really not looking to compete with the 70, 80, 100 companies now that are really trying to chase that, where I’ve got near monopoly or very few people in each of these verticals that I’m currently in. So I’ll keep it.
Dave Kruse: Yeah, well these companies would come and try to buy you, not that you would see, but…
Mel Torrie: Yeah, we are definitely getting offers every since we started. Only in the last 18 months to 2 years that we really had the private equity and venture folks calling weekly. They wouldn’t return my calls in 2008, but now it’s tough to get rid of them. So my thing is kind of like what Mark Cuban said, if you have an exit strategy, you are not passionate enough about your business. I have no desire to exit ever, and so that kind of money isn’t of interest. But strategic money, whether it’s for developing something that is of interest to them, whether it’s for equity or not, they are in it for the long haul, they are in it because they are passionate about being successful in that vertical and that’s the kind of money that I am interested in and so it’s great that the industry is getting the excitement, but we are definitely not looking to exist and trying to build a world leading company. So that’s the plan currently.
Dave Kruse: That’s inspiring and that’s a good way to end it, although I still need to ask about basketball. I think you listed – I forget what you put, it was one of your bios I saw basketball. So did you grow up playing basketball or what…
Mel Torrie: Yeah, I am fanatic. Again I was born in the middle of nowhere and so a high school of 37 kids, I got to start on the basketball team and so if you could walk you made it. And so ever since then I’ve played three to four times a week. So I’m still shuffling around it. How old am I? 46. I still shuffle around, get up at 4.30, 4.40 and go in and play with a bunch of old guys at school and then I’m a Utah Jazz fan and follow them pretty closely. But I just love the sport, love the team elements of it and I hope to play it till I die.
Dave Kruse: Yeah, well that’s healthy. I’m impressed that you play that much. That’s awesome.
Mel Torrie: Yeah, I’ve got tons of different braces on to keep me from falling apart and we try to keep our feet low to the ground and shuffle safely, but I definitely enjoy it.
Dave Kruse: But you need to develop an exoskeleton so that accidently…
Mel Torrie: Yes, I’m about due to start using one, that’s for sure.
Dave Kruse: Well, Mel I definitely appreciate you taking the time chat and this has been, like I said, inspiring, and I love that you created this without taking outside cash for equity and you figured out how the make it work and make it work in a big way and yeah. I mean, when you started this, did you think that autonomous vehicles will be this big 16 years later.
Mel Torrie: Well, I was probably a naïve optimist and believed it should have happened a lot sooner, but yeah, I think finally the demand is there. We’ve got – either Trump is going to set down the borders or Hillary is going to raise the memo and that’s just pushing automation all the faster. The demand seems to be converging with the cost of the technology and so it’s a very exciting time. So it’s kind of finally we’re right there versus some of the destination thinking where I was getting a little frustrated it was taking so long to get OEMs comfortable with the liability risk and getting the price points where people could afford it. So it’s an exciting time. We’re thrilled to be where we’re at.
Dave Kruse: Yeah, and your young. You got many years to make this happen, even on a bigger scale, which is cool. All right, well like I said, I definitely appreciate it and yeah, what you built is pretty impressive. So thanks for sharing your story and teaching us all about what you do and what you’ve learned over the years.
Mel Torrie: Well, I’m definitely honored to be a part of this show. Stay in touch. We’d love to keep you updated as we move into some of the high quantities of the wannabes in the market, so.
Dave Kruse: Yeah, that would be excellent. Yeah, and so Mel thanks again and thanks to everyone for listening to another episode of Flyover Labs. I hope you enjoyed it as much as I did. So thanks everyone, thanks Mel.
Mel Torrie: Take Care. Bye-bye.