Autonomous trucks are already being tested on select routes today, and it won’t be long before highways—and maintenance bays—are full of them. As with any disruption to the status quo, there are questions; and there is no exception in the case of commercial vehicle technicians curious as to how their profession will change to adjust to the maintenance of autonomous vehicles (AVs).
To understand what a maintenance team can expect when an AV rolls into their bay, Fleet Maintenance spoke with Dan Goff, head of policy for Kodiak Robotics, an autonomous technology company focused on long-haul trucking.
Fleet Maintenance: Can you explain what level of autonomy Kodiak vehicles operate at?
Dan Goff: At Kodiak, we are developing SAE Level 4 autonomous trucks. Level 4 basically means that a vehicle can operate without human intervention or human oversight, or even can be completely driverless under some relatively narrow circumstances. For example, you could have an SAE Level 4 vehicle that drops people off at their cars after a sporting event.
We’re focused on initially deploying our trucks without a driver on highway routes, and then expanding from there to eventually get from distribution center to distribution center. The technology is considered Level 4 because it can operate without a human. But it can’t go everywhere, and we are not planning on backing up to the local grocery store—at least not for a very long time.
Today, I think it’s important to note we are still testing these vehicles. There is one autonomous vehicle joke—you can tell a vehicle drives itself because it has two people in the cab at all times. While we are testing, we have a highly trained truck driver with a commercial driver’s license who is ready to take control should it be necessary. We also have an engineer in the passenger seat who is watching the computer, watching the system, and making sure it is behaving the way it should, and communicating with the driver accordingly. Driver-out deployment [in reference to no driver in the cab]—at least in some narrow circumstances—is something that we expect to see by the middle of the decade. It’s only a few years away. This technology is coming sooner than most people think.
FM: It is certainly exciting to overhaul the way that this industry has been operating for decades. Like you said, such digital technologies are coming down the road, but that horizon is a lot closer than people are anticipating. Can you dive into the technology suite Kodiak utilizes?
DG: First, it is important to note, we—and most, if not everybody in the industry—are building our technology on top of traditional trucking platforms. Our fleet is mostly Kenworth T680s. Our trucks use a combination of cameras, radar, and lidar, which is basically a laser system that sends out a light pulse and sees what comes back. Lidar can build a really detailed 3D map of the world around the truck.
We think it’s important to use a wide range of sensor technology, because each kind of sensor has its own pluses and minuses. Lidar is really good at creating this 3D map, but its performance isn’t as strong in inclement weather; whereas radar has a much lower resolution than a lidar unit, but performs really well in the rain and other bad weather conditions. In addition to those sensors, which get a lot of the attention, there are other things happening in the truck. We add and work with Tier 1 partners to add computer-controllable components—brakes, steering, things like that. We add a very powerful computer to the truck.
We also pull out the sleeper so we can add a couple of jump seats for engineer ridealongs and things like that. But mostly what you see on the truck is an off-the-line Kenworth T680.
FM: In terms of operating over an extended period of time during this development phase, those sensor components have to be maintained themselves. Are there recalibrations or software updates, things of that nature, to keep in mind?
DG: A big part of keeping these autonomous fleets operating are maintenance procedures. We’ve built internal maintenance processes working with our suppliers and understanding their best practices and their recommendations to build maintenance processes that take into account both the needs of the underlying truck platform, as well as the sensors and other components. Sensors do need to be calibrated, sensors need to be cleaned, and there are a lot of these operational maintenance requirements that go above and beyond what traditional fleets have to do.
Today, most of those maintenance processes are proprietary, just because the technology is a little bit different from developer to developer. As the technology develops, those will start to be a bit more standardized. One of the implications of that is that when we start operating driverlessly, we’re not going to be in a position in this industry to hand a truck over to a carrier and say, “Best of luck with this.” We are going to need to be closely involved in training the maintenance team at our carrier partners, probably even embedding some of our own people, to make sure that the trucks are in as good of a condition as they can be in and good enough to be on the road.
That’s an important point; we understand that we need to hold ourselves to a really high standard in this industry to gain that public trust. We understand that there is obviously going to be some skepticism about this. We’re making sure that the trucks that we put on the road are among the best maintained of any vehicles out there. It’s something we’re really proud of. You can see it when you look at the trucks—our trucks are maintained to an extremely high standard.
FM: There has to be the understanding that adopting these technologies will take partnerships—a blending of the best of both worlds. Maybe people are skeptical or hesitant, but people are also champing at the bit to adopt technologies that can really change the way the trucking industry is working.
DG: Absolutely, that’s right; and that’s true at a lot of levels. We see a lot of interest from regulators and law enforcement as well in the opportunity here to raise the level of maintenance and safety in the industry.
I chair the American Trucking Associations’ Technology and Maintenance Council’s Automated Commercial Vehicle Inspection and Enforcement Task Force. The task force is focused on developing commercial vehicle inspection processes for autonomous trucks. The processes we’re working on suggests “no-fault” inspections—where we have an in-depth pre-trip inspection, and the expectation will be that everything is spotless before a vehicle ever leaves the facility.
In addition, you have all these sensor technologies and analytics and telematics that can also tell you a lot about what is happening on the vehicle, and make sure that it’s safe, and help you keep things maintained the way they should be, as well.
There’s a tremendous opportunity here for these technologies to transform how maintenance gets done and how we keep trucks safe. For someone working in the maintenance industry, that’s really exciting because in addition to giving more certainty and more information, this means that maintenance becomes an even more important part of how the trucking industry works. It becomes the core operational question, “How do you keep these trucks maintained at the level that they need to be?”
FM: It is no secret that monitoring the vehicle through telematics creates a level of insight that isn’t available without it. How does Kodiak Robotics leverage telematics and remote monitoring? Particularly, following the recent investment in the company from Bridgestone, can you expand on what such technology offers Kodiak from an operational perspective?
DG: We use a variety of different telematics platforms, including off-the-shelf. We also have developed our own system that allows us to remotely monitor what’s happening on the truck and gives us a level of detail that you’d expect for autonomous trucks; it really goes beyond what is common in the industry.
In the long run, having access to those kinds of telematics platforms are critical. First, to maintain the safety of the vehicle, we’ve implemented Bridgestone’s tire pressure monitoring systems, and those are important to making sure that our tires are operating the way they should be on the road and allow us to take corrective action if they’re not.
These sensors also are going to have a real impact on our maintenance processes to go beyond just monitoring. One of the things we’re excited about working on with Bridgestone is connecting the data from their sensors to the information we collect about how our vehicle is behaving on the road, which we hope will allow us to get much more detailed information about what actually causes tire wear and tear, what kinds of pressures cause tire breakdowns and cause other issues with tires. Being able to connect those two datasets, we think, will allow Bridgestone to build tires better suited to the needs of autonomous trucks, and will also help us reduce wear and tear on the tires and do a better job of doing preventive maintenance.
These sensors are all critical to us understanding the operations of the vehicles, but they also will allow us to do all sorts of things to reduce our spend and reduce the costs of maintenance.
FM: From a maintenance perspective—having absolute predictive maintenance and knowing exactly how one system’s operation is going to impact the wear and tear on another—that’s the ultimate goal, isn’t it?
DG: A lot of this is going to benefit traditional fleets as well. On the maintenance side of things, a lot of what we’re talking about is not unique to autonomous trucks. We’re leaning a little forward on some things because we need to be. We need to have a high level of certainty about how tires are behaving on the road. But the benefits, and what we learn from this, are going to be pretty broad. Most of these technologies are not unique to autonomy. We are using traditional trucking platforms. These telematics systems are available in the market. That’s going to be really exciting for all kinds of fleets to see what they can learn.
FM: As Kodiak is monitoring its autonomous operations, has it shed any light on the implications autonomous operation has on wear items on a vehicle?
DG: For the most part, our system is a smoother driver than a human, which means less wear and tear. Generally, our system doesn’t break as hard or as aggressively; it accelerates more smoothly. That tends to lean towards reducing wear and tear.
On the flip side—today, we have a safety driver in the vehicle at all times, but when we begin to operate driverlessly, the goal will be to use the asset as intensively as possible. We’re aiming to operate 24/7, minus time for swapping trailers and fueling the truck. That is going to create some new maintenance requirements and changes as well. As we get closer to driverless, we’re going to be looking to simulate the environment of as close to 24/7 driving as possible. To some extent there’s some knowledge available already from team driving operations, but there’s definitely a lot that we’re going to need to work on if we get to that point where we’re deploying [driverlessly].
The wear and tear per mile will probably be less than on a traditional truck. But given the intensity of driving, I think the overall impact is that the trucks will probably need to be maintained more frequently given how many hours a day they’re spending on the road.
We have spent a lot of time thinking about that.
FM: What will be the major implications for the maintenance department with autonomous vehicles?
DG: To some extent, not a lot is going to change other than the requirement and the standard that trucks need to be maintained to. Generally speaking, 90% of what you’re working on, on the truck, is the same. The biggest changes are going to come from some new components. Again, the processes around maintaining those are going to be proprietary for a while, but I think over the next 10 to 15 years, there will start to be best practices and start to be more of a general industry understanding of how these technologies work and how to maintain sensors.
Another big thing to think about is geography. With maintenance taking the central role in keeping these trucks on the road, people are going to have to understand how these sensor technologies work in a wide variety of geographies. We can’t always be sure that a Kodiak technician is going to be able to get to a Kodiak truck when it might be that the truck is operating hundreds of miles from the nearest Kodiak facilities. So, maintenance organizations are going to have to figure out how to partner.
FM: Do you foresee anything with the traditional vehicle architecture becoming a higher-priority maintenance item than it is without autonomous technology specified to it?
DG: In general, keeping sensors clean and calibrated is a genuine challenge. We want to keep the sensors at a place where mud isn’t going to cover them because that would be a real problem. But it’s something that we have to think about. And, as part of those maintenance operations, having those sensors checked and checked well every time a truck leaves the facility is a part of that.
As there’s more technology on the truck, that technology tends to focus on the higher “risk” items. Maybe that’s the wrong word, but we will know with a great degree of certainty what the tires are doing, what the engine and brakes are doing, things like that. Where it gets more difficult is things like fenders, things that are hard to sense for. That’s where [we need to be] diligent from a maintenance perspective, making sure that the fenders are in good shape and those sorts of things.
We’re looking at ways to try to sense some of those things, but there are things that humans would notice immediately that are going to be much harder for an autonomous truck to notice, which is part of why preventive maintenance is so important. Those things will come to the fore a little more as more technology gets on the truck.
It is going to be a really exciting time for maintenance where these vehicles are on the road and it’s going to be so critical to operations—making sure that vehicles are maintained to a very high standard; it’s going to be one of the most interesting parts of this, figuring out that operations puzzle.
This interview has been edited for clarity and length.
