Caterpillar CTO: Technology Strategy, Autonomous Vehicles, and More | CXOTalk # 807

Caterpillar CTO: Technology Strategy, Autonomous Vehicles, and More | CXOTalk # 807

Show Video

Today on Episode #807 of CXOTalk, we're  discussing the intersection of business   strategy and technology with Karl Weiss. He's  the chief technology officer of Caterpillar.  We are the largest manufacturer of construction  and mining equipment, off-highway diesel and gas   engines, and the leader in gas turbines and  electro-diesel locomotives. We've got almost   110,000 people and, last year, we had revenue of  almost $60 billion. So, yeah, a large company.  We have over four million assets actively running  around the world today, and 1.4 million of those   are connected; connected to us, connected to  our customers and our dealers. Speaking of our   dealers, we have global coverage; 156 dealers  in almost 200 countries around the world. 

That's incredible. You have 110,000 employees,  109,000, and 1.4 million connected devices.   You're CTO of this enormous landscape. How do  you manage that? How do the pieces fit together?  My role as CTO is really two-fold. One is to  bring those folks together at a leadership level,   coordinate and align our technology investment  and platforming, and ensuring we serve our   customers consistently and get the most out of  our research and development investment. Secondly,  

I have some direct responsibilities that  surround our software development team,   all of our electronics, our autonomy and  automation team, our hydraulics systems   development team, and then the manufacturing  locations around the world that make our advanced   components like hydraulics, transmissions,  drivetrain, operator cabs, et cetera.  As you look across this broad set of  products and technologies, is there   a common thread or a common base that you use? If you asked our product development community   in particular that question, I think, by and  large, they will all answer, "It starts with   the customer." We are a very customer-led R&D  organization, meaning we don't develop technology   because we can or for the sake of showing we  can, but using the latest technology to help   our customers solve their greatest problems and  become more successful using our equipment and   solutions than with any of our competitors. That's really interesting when you talk about   being customer-led. How does that feed into the  business strategy? In other words, you've got the   technology piece. You have the business piece.  It seems like the glue maybe is the customer. 

If we serve the customer better than  anyone else, we will have good business   outcomes. If we develop those technologies and  solutions in a way that leverages our scale,   then we will have good business outcomes. I  think, using that customer as our North Star   kind of solves all of those equations at once. Can we jump into a discussion of some of your   products because you build this large  equipment? I don't think people realize   the extent of AI, machine learning, and  autonomous operation that's involved.  It shocks people sometimes to realize how  technologically advanced our organization   and our equipment is. Just to give you  an example, we have the largest fleet of   autonomous mining vehicles in the world. We've  been developing it for a couple of decades but  

have been in a production environment for  over ten years in mining around the world.  Today, that fleet of almost 600 trucks travel  the equivalent of more than 3 times around the   Earth every day without an operator in the  truck. So, it's pretty amazing technology.  What's the evolution or the history of  that kind of technology at Caterpillar?  We started this journey with the DARPA Challenge  (a couple of decades ago almost) and really   leveraged university talent and our own talent  to develop a driverless truck. Maybe making the   truck a robot was the easy part. The hard part is  then making it productive and having a fleet work  

together autonomously in an operating environment  that has dust, obstructions, and obstacles.  When we began this journey and first put our  trucks in a mine site, we had more than 200   stoppages every day because of different reasons  just to make sure everything was safe. Over time,   we gather a lot of information through our LiDAR,  through cameras, and different ways of sensing   the environment. We've been able to use machine  learning to really teach our algorithms (over   time) how to adjust to the environment, what to be  concerned about, what not to be concerned about,   how to approach different obstacles in  different ways. Through that, I think  

over 230 million kilometers of travel, we have had  zero injuries. So, we've been very proud of that.  The safety aspect, maybe weave that  into it because, unlike many products,   in your case if there is a product failure, it  can literally lead to death or severe injury.  Our largest mining truck carries over  400 tons of material, so it's a very   large vehicle. And safety is one of the  reasons that our customers have moved   toward autonomous vehicles in their mine sites. We have seen, and our customers have told us,   that their injury frequency, their severity of  injuries have declined in all the cases where   we have implemented autonomous vehicles. We  are proud of that and continue to offer our   customers even more improvements in  that space to ensure that continues. 

Why is that? What is it about the autonomous  vehicles that have such an impact on safety?  A lot of our mine sites are 24-hour operations.  That takes its toll on human operators.  One of our first safety devices was the ability  to alert the driver and their management of   fatigue when drivers would fall asleep. And so,  that all by itself has been an improvement in,   first, helping them understand fatigue and  how to avoid it and to alert people when   they are fatigued. But that has moved to the  ability to then take the operator out of that  

equation and not have that risk. There are other risks, of course,   in the mine site. That's just one good  example. But our vehicles are able to   just consistently continue the operation they've  been asked to do without distraction of fatigue.   I think that that's made a large difference. Please subscribe to our newsletter. Subscribe to  

our YouTube channel. Check out  We really have great shows coming up.  We have a very technology-focused audience. Can  you give us a sense of the core technologies that   make autonomous vehicles at Caterpillar possible? LiDAR has been a key technology for us from the   beginning, and the ability to look far enough  in the distance and through a lot of different   environmental distractions has been critical. The  ability to capture millions of points per minute   has given us the ability to have confidence  of avoiding obstacles and staying on course.  As we've evolved, we're using more and more  camera technology and radar technology for   different purposes to also watch different  parts of the vehicle so that we can avoid   other safety incidents that may not be straight  ahead but beside the vehicle or in back. But   all of this is dependent on connectivity to  GNSS (satellites) and knowing where we are. 

As we evolve, we are working toward a more  local-based knowledge of the vehicle at what's   around it so that we can improve our capabilities  even when we lose line of sight to a satellite,   as an example. It's been a journey,  and we continue to improve upon that.  That's really fascinating; millions  of points of data per minute.  Yes. It's an extraordinary  

amount of data, and I assume gets to the earlier  point that you made right at the start that you   have 1.4 million connected devices out there. It's a scalable strategy when we talk about   connected assets from connecting a very small  piece of equipment for a lawn and garden care   provider to this mining truck that is operating  in an autonomous environment. The amount of   data per minute, per hour, or per day is very  different, and so we're able to serve both ends   of that spectrum with various levels of data. For instance, in our mass amount of product, we   are operating really in a cloud-native environment  of where to store and process our data to this   mine site where this massive amount of data has  to be real-time (for obvious reasons around safety   and performance). Those are on-premises data  storage, and so we have a great team that does  

an excellent job of looking at the different  needs of our customers along that spectrum.  The whole strategy of data  and analytics seems to have   transformed the nature of Caterpillar's business. For sure. We began our telematics journey really   in the 1990s, but it was more of a niche part of  our business where we engineers were working on   how to get better, faster access to information  when we test in the field. As capabilities have   improved, the ability to send and receive  data in mass amounts at a reasonable price,   and the computing capability to scan all of that  data and make it useful, has transformed how we   treat our data and how we have now standardized  on putting telematics on all of our equipment   as it leaves the factory so that we're able to  help our customers be safer, more productive,   and save money all at the same time. In a way, there are – correct me if I'm  

wrong – two broad parts to your technology. One is  the underlying machine, the physical machine that   you're building (including the engines and all of  those physical components). And then you have the   data analytics, telematics, telemetry part of the  company that involves machine learning. These are  

very, very separate kinds of technology tracks. Yes and no. They come together in many ways   because, for instance, the transmission has  several sensors in it, and we have the ability to   take the data that we've collected over that 1.4  million assets and scan the data for anomalies to   then be able to more proactively let our customer  know that a problem may be occurring. Whether it's   an operator operating in an unsafe way or whether  it's a clutch that's going to wear out or in an   engine when the oil gets diluted to the point that  it needs to be changed, we're able to catch those   and speed up the information to our customer to  save them downtime and a costly repair. They kind   of come together in how we're treating the data. What about from a machine learning perspective?   Where does that come into play? I'm asking because  that's what everybody is interested in right now.  

So many organizations are looking at AI  and looking at machine learning to figure   out how to improve whether it's their internal  operations or improve their products. It seems   like this is a core part of your innovation. It certainly is. Maybe I'll give you a   couple of examples to give a flavor for it. In our autonomous and automated operations,  

we're able to gather that information that's  coming in at a very rapid rate at a massive   amount of data and teach the algorithm, over time,  to reduce the number of stoppages in the operation   so that, over time, that system becomes more  robust. That learning is just an ongoing process   and it makes us better and better over time. Another area I would talk about is many customers   ask our dealers and ourselves to help them  manage their fleet in looking at a larger fleet   of equipment, looking for those anomalies, and  we have fleet managers that sit behind a screen   looking at fleets of machines and looking for  those anomalies, the fault codes, temperatures,   and different things. We've been able to use  machine learning to speed that process up to  

make one fleet manager much more productive  and to provide much quicker feedback to the   customers and dealers on where they should  focus their time because we're all short of   talented people right now and this just enables  our teams collectively to be more productive   and to save our customers time and money. What about the data architecture that you've   put into place to handle this level of scale? We are managing that data largely. I mentioned   the on-prem data is a little bit different  than the rest of the 1.4 million connected  

assets. But for the large amount of  equipment around the world, we have a   cloud-native environment where the gateway that  connects to our equipment, the data processing,   and the storage all happen in the cloud. Having that cloud-native environment   (kind of serverless environment for us) really  allows us to maximize the use of our investment   and really minimize the overhead costs long  term for us. And it allows us to keep up with   the latest and greatest speeds and storage  capacities that the industry has to offer. 

What about electric vehicles? We have very small equipment that   is used (lighter-duty applications) a few times a  day in urban environments to the very large mining   equipment that we talked about earlier. Both  ends of our equipment line – and I would put   data center support in that large end of things –  are moving fairly rapidly toward electrification.  The small end because we can leverage automotive  scale and they're in environments that allow for   longer operating times on battery. And to the  very large end because our mining customers   and our data center customers are very focused  on sustainability, and they want to be able to,   let's say in a mine site, operate with  a lower impact on the environment,   and they've made commitments publicly about  getting their mine sites to that level.  Last fall, actually, in 2022, we introduced  our 240-ton payload truck that was fully   electric drive, operated from a battery. We  demonstrated it at our Tucson proving grounds  

and showed that we're able to make a full-load  production trip and operate as well as our   diesel equipment in that environment. In the middle of that equipment line,   that will take longer for reasons of access to  the grid, for reasons of the ability to run all   day with a charge. In a lot of locations around  the world, they're asking us to really look at   alternative fuels that will lower their impact on  the environment while enabling them to be mobile   and be productive without a grid nearby,  so we've got a lot going on in this space.  This push for electric vehicles, is that  being driven primarily by sustainability   or vehicle performance or what? What  are the underlying drivers for that?  It's really a customer-driven need, and  we take the lead from our customers on   what they need to be successful. I think, in the very small end of  

our equipment range, we have customers  in environments that are really going   to necessitate the need for electric vehicles.  They are genuinely interested in reducing their   environmental impact and want us to provide  them the best option that we can to do that.  Then on the far end, on the mining side, we're  there directly supporting our mining customers   in their commitment to getting to a carbon neutral  position at some point in the future (depending on   the customer). We're very actively engaged with  them developing that mining equipment to do that. 

Your customers are (for a variety of reasons)  requesting electric vehicles. And so,   that pushes you to develop the R&D  that's needed to accomplish that.  That's right. We've made commitments ourselves  relative to sustainability. For instance,   for every new or improved product that we have  in our portfolio, we have committed that all   of those that will be introduced between  now and 2030 will continuously improve on   reducing environmental impact that they have on  the ecosystem as we do our product development.   It's really both driven by our customers  and our own commitments to sustainability.  We have a couple of questions from Twitter. Why  don't we jump to those? Our first question is from  

Chris Peterson who is asking a moonshot question.  Well, literally, a moonshot question. He's saying,   "Is Caterpillar planning to play a role in  autonomous or remotely operated machines   for possible use on the moon or Mars in the  future because bases will need excavating?"  I've been lucky enough to spend time with  NASA in the last few years because we actually   have sponsored a Lunabotics competition  with NASA over the last several years.   I've been to the Kennedy Space Center,  been to their labs where they replicate   (as best we can) the regolith, the moon soil. Our competition has been for university teams to   develop a small machine that will remote-control  or autonomously operate in that regolith and be   able to drill down through the regolith to extract  rock because that's a very real application that   NASA is interested in. We've been working  with them and helping them on that journey.  We have another really interesting question  from Twitter. This is a completely different   subject from Lisbeth Shaw who says, "What kind  of innovation (either product innovation or   business model innovation) is part of your set of  responsibilities or your remit at Caterpillar?"  My team is responsible for what we call the new  technology implementation process and the new   product implementation process. And so, we're  constantly working to lean those processes out  

within Caterpillar, and we set the process, the  guidelines, and standards for the company on how   they run those product development programs. In that way, we're consistently working to   make improvements on our business processes.  We spend almost $2 billion a year in R&D,   and so that's a very big deal for us. Relative to innovation on product, we're   really at the heart of the main componentry of  Caterpillar. My direct team does all the software   development, so we are working to speed up that  process and take errors out of that process.  One of the most important parts of our product  development time is testing. When I was a young  

engineer, we always designed. We'd build a  prototype. We tested it. Then we built a pilot   machine to run down the manufacturing  line. Then ran it in operator sites.  Today, we still do some of that, but a large  part of what we do, we use what I'll call   digital twins of components that we build in the  software environment and run tests, whether it's   finite element analysis on structures, whether  it's cooling analysis on our cooling system,   or whether it's performance and thermal analysis  on transmissions. But we do much of that now   in a computer environment, which speeds up  the time it takes us to develop a product. 

We're able to run our systems through a hardware  in the loop and software in the loop to test   out our software and to do it in time much  faster than real-time (in many cases). So,   we're kind of at the heart of a lot  of those improvements to our business.  That's interesting. Digital twins are  really, it sounds, integral to your   technology and product development. Definitely. More and more integral.  What about the relationship between  business strategy and technology strategy?   You are a chief technology officer, which of  course implies the technology aspect. But where   do you intersect the ultimate business  planning and strategy for the company?  Every year, we organize a group of senior  technologists. We call it the Senior Technology  

Leadership Forum. There are chief engineers and  leaders across our different business units.  They come together, and they debate and vote with  their money on where we invest our future dollars,   our collective research dollars. Then they make  a recommendation and bring it to our Product   Development Council, which I chair (and includes  my peers across the many business units) to   really fund those collective investments, which  really leverages the scale of Caterpillar and   also drives consistency in the technologies that  our customers eventually see in the marketplace.  The other way is I talked about our new product  development process (NPI process). We have set   up the framework that enables us to measure  all of those product development projects and   how much investment is approved and making  sure that those development programs meet   the goals (whether it's performance, quality,  cost, and timing) to ensure that we're getting   the most out of our R&D investments. What are the key performance indicators  

(KPIs) or metrics that you use for  evaluating things like R&D investment?  We do compare programs of similar nature to each  other and are continuously looking for ways to   be more efficient. But really, it comes down  to performance, quality, cost, and timeline,   and ensuring that there is an alignment (before  the program starts) between the technologists,   the program managers, and upper management  on what the goals are, how much will it cost,   the timeline, and the expected return.  Then we track those over several years.  How do you drive that alignment, because  somehow I'm sure it's not simply, "Oh,   let's have a cup of coffee, we'll figure  this out, and then we'll go on our way"?  There are many tough discussions and meetings  along the way. I'm very proud of the fact that   we have a deep bench strength of experts  that have done this a few times and are   able to really collaborate for what we think  is the best for our customers and for the   enterprise. But that doesn't mean we always agree. It takes us a while, at times, to get to that,  

I'll call it, consensus. It's not  always, but yeah, we aim to get to   a consensus on those big investments, for sure. Aligning, as you just said, what is best for the   customer along with what works for Caterpillar,  and then I'm sure another layer on that is what is   practical and feasible from a technology point of  view within whatever the reasonable timeframe is. 

Exactly. There is always more that we  would like to do than we can afford or   have time for or resources for. We have a  lot of customer requests and expectations,   and our teams always want to meet those as fast  and as best as they can. We can't do everything  

for everyone, and so the key part of all that  is really prioritizing and saying where will we   make the most impact for our customers and where  does Caterpillar have the advantages that we can   leverage so that we're serving them best. We have another question again from Chris   Peterson – another thought-provoking  question totally unexpected for me,   as unexpected as the moonshot question – which  is, "Do the digital twins and software simulations   roll forward into AR and virtual reality tools for  teaching and doing maintenance and operations?"  It's a journey. We have a team focused on AR and  VR because it's becoming more and more important   to our ecosystem and gives us capabilities  of doing things we couldn't do before.  As an example, it's very important for us, as  we develop a new product, that it is repairable,   that it's easy to do maintenance on it,  and that it's safe for our dealers and   customers to work on that equipment. It used to be, when I was young,  

that we built those products and, for the first  time, had our own mechanics work on it and give   us that feedback. And we would do a safety  audit and then, after you built the first one,   you had to go back and make the changes needed. 900:34:25) Whereas now, using a virtual reality   environment, we're able to do much of that, most  of that, really virtually with those mechanics now   where they can put the VR headset on. They can try  to get inside the equipment virtually and reach,   and we can do a high percentage of our safety  audits that way. That's just one example of how   we're using that technology. That technology is actually  

in practical use today, AR and VR? Yes. In fact, we used to have what we call   a cave that we've probably had for almost 20 years  that was very much a projection of the equipment   around you. Whereas now with the VR technology,  we're able to put the headset on and live that   environment without being an actual cave. You mentioned safety earlier. Given the   huge importance of safety, how do  you think about that and manage   that? Again, I'm sorry, very quickly, though. Safety is of huge importance to us and Caterpillar   has consistently improved our reportable  injury frequency to world-class levels. Today,  

we're very focused on ensuring that the severity  of any injury is minimized and eliminated, and   so we're focused on making sure folks in our own  organization are outside of any line of fire or   risk. And we're looking at the same things for our  customers that operate and maintain our equipment.  How does technology and innovation strategy  get addressed in the system architecture?  We want the system, especially the electronics  and information system, to be upgradeable very   rapidly, so we are developing architecture,  software architecture, that's agnostic to   what display you use, how you use it, and we're  able to use more of a central domain computer   that has a partitioned architecture to allow that  modularity and upgradeability. For a short answer,   that's where I'll leave it. The modularity is independent   of the specific product because you  have underlying components that I'm   assuming are then adapted to bigger, smaller. Right. Exactly. More plug-and-play capability.  International, you're in a very broad set of  markets, and you've even spoken about excavations   on the moon. So, how do you adapt your technology  strategies to meet these very diverse geographies? 

It's really important for us to do what  you just said. I would say there are three   things we do to make that possible. One is we really pay attention,   working with our dealers, on what the local  customers need because they're different   and they have different economics in some cases. Second, we work to hire the best talent in-region   because they know that environment. They know  the customers. They've lived it. And if we   get the best talent there, we know we will  develop the best products for that region.  Then third, we work closely with governmental and  non-governmental agencies to make sure we have   common sense standards and laws in those areas  that both support the customer and our business. 

What about your global supply chain? How do you  ensure supply chain resilience given the size,   the diversity of your products, and so forth? We have, like most companies, worked on a   just-in-time lean supply chain, and that's  bitten us in the last few years with the   pandemic, the supply chain disruptions. One of the things that have been really   helpful to us is that we tend to build products  in-region for the region. We don't always do that,   depending on the volume. But that's  how we've set up our manufacturing   and supply network, so that has served us well. However, we have paid more attention more recently   in ensuring that we have more resiliency, maybe  more multiple sources rather than single sourcing   certain components. Of course, microchips  was a big part of our supply constraint in  

the past upturn with the pandemic as well, and so  we've taken steps through sourcing and inventory   buffering and working with our tier one, tier  two, and tier three suppliers to improve that.  What technologies are you excited about  and where is the set of technologies that   you're working with headed right now? I've been an engineer and at Caterpillar   for 35 years, and this is definitely the  most exciting time of my career relative   to what's possible. What's really important is  the combination of some of these technologies.  We've talked about, generally, all of the  key ones, but we call it AACE technologies   (autonomy, alternative fuels, connectivity, and  electrification). How we bring those together,   which we talked about in most of this  hour, is whoever does that best and   makes it easiest for a customer will  win and will help them be successful. 

With that, it seems like a good place to  end. We are out of time. I want to say a   huge thank you to Karl Weiss, Chief Technology  Officer of Caterpillar. Karl, thank you so,   so much for taking your time to be with  us today. I really, really appreciate it.  Thank you, Michael. It was a  very engaging conversation.  Thank you to everybody who watched, especially to  those folks who asked all those great questions. 

Now, before you go, please subscribe to  our newsletter. Subscribe to our YouTube   channel. Check out We  really have great shows coming up.  We will see you next time.  Have a great day, everybody.

2023-10-21 03:17

Show Video

Other news