Last month’s Manifest conference in Las Vegas aimed at bringing together a spectrum of global supply chain executives, logistics service providers, innovators, and investors at the forefront of logistics tech. This fourth iteration of the conference attracted over 6,000 attendees from around the world.
Automation for every imaginable human task in logistics was represented, both indoor and outdoor. Among 100 or so exhibitors, there were three focused on autonomous trucks: Forterra, Bot Auto, and Torc Robotics . Other companies in the AV trucking space sent delegates, including Fernride, HubPilot, ISEE, Plus, Outrider, Stack, and Waabi.
The first Manifest in 2022 saw intense interest from virtually every truck autonomy player, with the back of the exhibit hall showing a long line of prototype robo-trucks. Several of the companies have left the autonomy space by acquisition, re-focusing, and/or, in some cases, bankruptcy. TuSimple and Embark are no longer in the mix, while Einride is focusing on both EV trucking and autonomy. Between then and now, the number of AV companies in the exhibit hall has gradually reduced, even while the promise of launching autonomous operations comes more into focus. This is not surprising. As an Automated Driving System developer moves closer to launch, there is less need to seek attention because the company is already getting lots of attention from their deployment plans or – in the case of Kodiak – their first launch of commercial driverless operations in an off-road environment.
Of the dozens of sessions at Manifest, two panels focused directly on truck autonomy. The speakers, longtime leaders in their fields, provided intriguing viewpoints on how autonomy fits into the on-road aspects of supply chain operations.
The first panel, titled “Autonomous Trucking: On the Cusp of Commercial Driverless Operations,” was moderated by myself and consisted of Peter Schmidt, CEO Torc Robotics; Josh Araujo, CEO, Forterra; Paul Enos, CEO, Nevada Trucking Association; and Dustin Koehl, COO, Covenant Logistics.
The second panel, titled “Accelerating Autonomous Solutions,” was moderated by Lee White of LM White Consulting and consisted of Bob Hall, COO Outrider; Xiaodi Hou, CEO and Founder BotAuto; and Hendrik Kramer, CEO and Co-Founder Fernride.
The ADS developers represented in these panels covered both over-the-road trucking and yard truck operations. More than a few of the comments were quite striking and insightful. I’ve amalgamated the key perspectives in the following sections.
Autonomous Over the Road Trucking
BotAuto is the newest player in the ADS trucking scene, coming out of stealth last year. Mr. Hou offered a fresh perspective on what this space is all about, saying “We are not a tech company, we are a trucking company. What matters is cost per mile. AV players must deliver capacity.” Compared to the early ADS developers, he projected a fast deployment cycle, saying BotAuto will launch driverless operations on the roads in Texas later this year. He emphasized that the development of a self-driving machine is much faster than in the past, because a “foundation” AI model can quickly learn how to take care of most of the driving. For BotAuto, using one AI model, “in one or two months we got a pretty amazing model.” That said, he cautioned that “I am a big antagonist about end-to-end AI. You should not rely on the neural nets for rules of the road and Newtonian physics; these need to be added separately as another layer.”
“Regarding AI, what’s different before and after ChatGPT? A new way has emerged. You don’t need millions of miles” to develop the self-driving capability, he said.
Mr. Hou elaborated further on his approach. “Once you do safety, what about efficiency of computing? We can use ‘inefficent but safe solutions’ for simpler situations, making sure we don’t compromise safety. We transform all safety problems into efficiency headaches. We’re growing our efficiency step by step. When our efficiency goes beyond human capability, that’s a good product.”
While BotAuto is in the early stages of bringing a product to market, Torc Robotics is at an advanced stage. In just two years from now, the company plans to enter the market via their OEM partner Daimler Trucks. “I am super excited about the next three years. It’s really all about software testing and operations to deliver a commercially ready autonomous platform,” said CEO Peter Schmidt. “Now, it’s really about building a product and building a real business, which is safe and which can scale.”
An audience member asked the question that is probably on the minds of a significant portion of any audience that I and these panelists ever addressed: “How can we trust autonomous vehicles to perform safely when our laptops, phones, and other tech are frequently having glitches?” Mr. Schmidt’s response was one of the most robust I’ve ever heard, emphasizing the need for AV’s to have extensive redundancy. “I can lose 50% of my entire sensor suite and it still can drive. It’s the same with compute,” he said. “So even if a compute function goes off, your brain is there, you still have 50% of computing power. And that’s a lot of redundancy in all systems, not only braking and steering. You need redundancy in your sensor suite, your algorithms, and your compute and that’s driving cost. These concerns are valid. You need to design it into your architecture early on; if you try later on, that’s tough.”
Autonomous Yard Tractors in Distribution Yards and Ports
Over the years, Manifest delegates have witnessed a procession of autonomous yard truck developers showing capability not only in driving but also in efficiently connecting / disconnecting trailer hoses to and from the tractor. These connections are essential in pulling a trailer in a logistics yard.
This space has matured significantly over the last couple of years and limited scale autonomous yard operations have been launched.
At this year’s Manifest exhibit space, Forterra displayed a Kalmar Ottawa T2 Terminal Tractor equipped with Forterra’s AutoDrive™ autonomous driving system. AutoDrive is currently deployed on more than a dozen vehicle platforms across industrial and defense verticals. The vehicle was connected to a trailer mockup, enabling Forterra to demonstrate end to end auto-coupling with technology partners Electrans and Stoughton.
Customers demand that the trailer-connect operation take no more time than a human does in doing the same job. The auto-coupling tech performed the task in a flash, to my eyes faster than even a seasoned truck driver could do. Check out a video of this operation here.
In the panel discussion, Forterra’s CEO Josh Araujo explained their multi-faceted approach to ground autonomy. “We are deploying autonomous systems for military applications, both combat and logistics environments. And then we’re also working in industrial applications, ports, and distribution centers. So we’re looking at use cases that are off highway, off road, in ruggedized environments, adapting military solutions to similar industrial applications that can benefit from the technology.”
I asked Mr. Araujo to further explain advantages of the combined military-commercial focus. “There’s some definite benefits to operating in both environments. There are mutual synergies in the data, the experience in the field, and running edge cases in the military that you’re definitely going to see on the commercial side and vice versa,” he said. “From the military side, you typically don’t have the type of competitive pressure in defense acquisitions that you do in the commercial market. The size of the commercial market results in cost-efficient approaches, therefore providing much lower cost and higher capability systems to war fighters.”
How does this impact supply chain more broadly? “When it comes to supply chain factors, we look at really unlocking that value of supply chain visibility and planning. Having full end to end connected distributed systems that are autonomy-enabled is really how you do it,” he said. “That’s where we’re focused: inside that dock, inside that warehouse, and inside of the distribution center, providing that connection between the on-road leg of the haul to getting that trailer to the correct dock or moving it from the dock to the manufacturing plant or moving it from the ship side over to the stack side of the containers.”
Founded in 2018, Outrider was one of the first players in yard truck autonomy. Mr. Hall said the company “aims to accelerate deployment by putting little mental load on companies. This means, from the beginning, minimize changes for a customer to adopt autonomy.” He continued, noting that “the most important thing in logistics is throughput. The logistics yard is like a heartbeat, pumping freight out the way the heart pushes blood. We want to implement automated vehicles piece by piece, not as a single heavy lift. Our trucks can be introduced one by one, working side by side with human operated trucks. Over time, the number of AV trucks can increase. A big capital investment all at once is not required.”
In response to a question about the impacts of AI, Mr. Hall described Outrider’s journey. “Based on years and years of pilot deployments in large facilities, Outrider has accumulated millions of hours of data. Now we seed this into propriety reinforcement learning models. The rate of improvement and learning has been amazing. Over the past year, we overhauled the Outrider stack to incorporate AI models. This has allowed us to speed up the process of learning to adapt to new environments on the east coast and in northern environments.”
Later, as the panel discussed safety aspects of autonomy, Mr. Hall noted that “in North America there are no safety standards for yard operations. We pulled from existing standards to create our own safety case for yards.” But to the customer, what is prove-able safety in the yard? “It’s reliable and boring industrial operations. Just like you don’t sit and watch your dishwasher or toaster operate, you want to turn it on and let it do its thing. You want to have proven that it’s safe, even when you’re not looking,” he replied.
Based in Munich, Fernride is the only automated yard truck developer addressing the European market. The company’s initial product deployments have utilized tele-operation of yard trucks. Autonomy is on the horizon. How are they using the new AI tools? Mr. Kramer noted that human-assisted autonomy can co-exist with full autonomy. “As much as AI is needed, 95% of operations can be done autonomously, with maybe five percent requiring teleoperation. Having both options provides maximum reliability and flexibility in the business case,” he said.
He framed Fernride’s near-term priorities as follows: “The focus now is proving the safety case, i.e. verified and validated, to obtain the compliance with the European regulations in 2025. The focus in the next few years shifts from technology to operations. It’s not only about day one problems, but how are systems maintained and serviced for the long run. The key is to create value for customers.”
Voice of the End User
Not surprisingly, the tech developers had plenty to say. How does all this activity add up when viewed through the eyes of end users, i.e. trucking fleets?
Dustin Koehl provided a long view based on his many years interacting with the autonomous truck developers. “No longer is automated trucking just a science experiment, the developers are getting closer to commercialization,” he said. Covenant Logistics operates 2,500 trucks providing expedited freight services. For their operations, I asked Mr. Koehl “what does it mean for a fleet to deploy autonomous? Nothing changes? Everything changes?”
“A lot changes,” he said. “I spent almost two years working with a developer, where I gained a perspective of the investments that are being made on that side. And based on 18 years in the fleet world, I bring the fleet perspective.” Mr. Koehl framed the space in this way: “There’s two business models in play here. So you have the ‘driver as a service model,’ in which the autonomy developer delivers the truck to the fleet and says, ‘Here you go, we’ll support you as needed.’ But then also you have the transportation as a service model.” With TaaS, a traditional fleet outsources freight to an ADS developer who is fully responsible for delivering the load.
He bottom-lined this by saying, “Covenant hauls anything from live haul to eggs to bombs and ammunition; because of this there’s just always going to be a need for a truck driver there. While we stay close to autonomy developments, we’ve seen LTL and freight forwarders, even drayage fleets, getting ready for this, spending significant money to optimize their networks for the autonomous future. We stay close to it because of the unit economics of autonomous operations. If you can get 16 to 18 hours of utility on an autonomous truck, we have a responsibility to really understand the benefits of that, not just for us, but for our customers too.”
For Paul Enos of the Nevada Trucking Association, “the most compelling thing about autonomous trucks is stopping accidents.” He painted a vivid picture of the crushing liability burdens that on-road freight carriers face. This is exacerbated by aggressive plaintiff attorneys, even in cases where the trucker’s rig wasn’t at fault. “Liability is always first and foremost in my brain,” said Mr. Enos. “How do you insure these things? Who is the company that is responsible for an accident that occurs with an autonomous truck? When we passed our regulations allowing autonomous trucks in Nevada, trial lawyers called for a $25M minimum insurance policy on autonomous trucks. So, I think all of this needs to be worked out.”
Building on Mr. Enos’s comments, Mr. Koehl pointed out what still needs to happen. “Quite frankly, in spite of the nuclear verdicts and the risk, some of the insurance companies don’t really have products formalized for autonomous trucking yet. We know that fleets are not going to take this risk and liability on their own.”
Labor Viewpoints
Even though the rise of software, expert systems, and the like have been making a wide range of jobs obsolete for decades, when it comes to autonomous trucking the outcry about the fate of truck drivers has been at a fever pitch. A range of viewpoints across both panels painted an intriguing picture.
Mr. Enos spoke to the trucking world of today. “One of the biggest issues that our member companies have is hiring qualified drivers. They say, ‘Hey, we’re putting these guys out on the road with an 80,000 pound missile.’ This kind of stuff keeps the CEO, the safety director, anybody whose head is on the line, it keeps ’em awake at night. So I think having these driver assisted technologies that are available now are going to provide that safety cushion that makes it better for our industry, that makes it better for the drivers, even enhancing training.”
The key question he raised was, “How do we make sure that the folks that are coming into our industry, that we do need in our industry today, are not looking at the technology necessarily as a replacement for drivers in every application, but as that safety cushion that is going to enhance everybody’s experience on road.”
Mr. Koelh noted that “you’ve got this trillion dollar marketplace of truckers which is very segmented, and that’s not going away. As challenging as the truck driver job is, autonomy is not going to eliminate a lot of jobs. I was responsible for a budget of $50M a year to attract truck drivers to come work for our fleet. Its not a glamorous job. These drivers sacrifice a lot, it’s one of the most challenging jobs in America.
Mr. Araujo provided an interesting contrast between truck driving in the military versus the civilian sector. “On the commercial side for autonomy, the pushback is around unions and labor issues. The military is going to experience this exact same thing. They’re going to recruit, and they already started to recruit fewer truck drivers. But they have this current workforce. What the military has right now is an organic captive vocational training system that’s probably second to none, where they can take an entire workforce of drivers and retrain them in technical skill sets that can either support autonomy or do different value-added things on the battlefield. So, instead of having somebody behind the wheel of a truck, you have somebody that’s maintaining that autonomous system or a mechanic or command and controlling fleets of vehicles. And so that’s something I think the commercial markets can look at and say, ‘I have this amazing workforce that I can now retrain to do different tasks as they adopt their operations around autonomy.’”
From the Outrider perspective, Mr. Hall sees humans and robots working side by side. “This is how AV gets deployed: enabling human teams. It is hard to find people who are willing to do these jobs. Our Forture 100 customers find it harder and harder to recruit and retain drivers.”
And from the European perspective, Mr. Kramer described their ongoing operations. “We successfully operate in unionized environments in Europe. Unions in ports are friendly toward automated vehicles, because they can’t keep up operations due to lack of workers. They need AV’s as fast as possible. We’ve been successful proving to customers the autonomous tech works; they’re ready to roll out the solutions.”
“This is happening.”
It’s clear from the discussions at Manifest that the coming years promise to be significant inflection points for commercial operations of autonomous trucks.
In contrast to discussions at Manifest in earlier years, there was not a single mention from these well-informed experts regarding whether the technology is ready. Instead, the clear vibe was “This is happening” and “We’re on the home stretch” for commercial operations of autonomous trucks. And for shippers, as Lee White put it, “Its coming and you’re gonna be impacted.”
Disclosure: Richard Bishop is an Advisor to and/or equity holder with the following companies mentioned in this article: Forterra, Plus, Outrider.
