Q&A: The State of Autonomous Vehicle Development

Trucks with varying levels of autonomous or self-driving technology have been with us for more than a decade, yet no company has yet placed a full Level-4-autonomous truck into regular revenue service without a human driver aboard.

In the early days, few predicted it would take this long for self-driving technology to hit the road. Daimler Trucks in 2014 showed of its vision of the future with its Future Truck 2025. Where is the autonomous/self-driving truck industry today? How much longer will we have to wait for Level 4 to mature? What's the holdup?

To answer these questions and more, HDT invited Joanna Buttler, head of Daimler Trucks’ Global Autonomous Technology Group, and Peter Vaughan Schmidt, CEO of Torc Robotics, to discuss the current state of heavy duty self-driving trucks on HDT Talks Trucking

Read highlights of the interview with HDT Equipment Editor Jim Park here (edited for clarity and length) or watch the full interview above.

HDT: You have said that autonomous trucks can reduce accidents, and that you hope one day to get to zero crashes and zero fatalities. I don't know how realistic that is, but what kinds of crashes do you hope or expect autonomous trucks will be able to reduce or eliminate?

Schmidt: Crashes related to speeding, distraction and fatigue can be avoided just through technology, because computers never get aggressive or tired or distracted.

The second category where the system, in principle, can do better than humans is reaction time. For example, the trucks drive incredibly well in strong sidewinds, even with gusts, because the system can react within milliseconds.

In situations like really inclement weather, snowstorms, etc., the computer might do worse than a human. To be honest, that’s why we’re not deploying the technology in places like Montreal or Chicago.

HDT: If you looked at a statistic like the number of crashes per 100,000 miles, and you compare human performance to the computer, I have to believe the crash rate for humans would be considerably higher. If that’s the case, why aren't we ready to go 100% on Level Five? If you've already established that autonomous drivers are safer than human drivers, why are we still waiting to take the driver out of the truck?

Schmidt: There’s still a lot of polishing that needs to take place. It’s not so much the standard driving scenarios, but all the edge cases where humans are super great. We have to make sure all the edge cases are solved to a human-like level, because as good as an exemplary human driver must be the minimum threshold. And it's a really high bar.

On the other hand, there is still a lot of basic development to be done to make a supercomputer network and an automotive water-cooled computer last a million miles.

HDT: Can you offer a 40,000-foot view of how autonomous driving technology has evolved from its earliest forms to where it is today?

Schmidt: The earliest forms go back to the famous DARPA (Defense Advanced Research Projects Agency) Urban Challenges of 2007.

Torc Robotics, now a 17-year-old company, came in third that year. Back then the technology consisted of some super expensive handmade sensors and a purely robotic approach. We were mainly driving off a map. We had some laser scanners as well to determine where we were and what people around the vehicle were doing.

Since then, Torc has worked in the mining and defense industries, developing and improving the technology. Today’s sensors are on a completely different level of performance, speed, density, and range. And the computing power that you get today at an affordable cost is also in a completely different range.

The AI machine learning back then was more or less classical algorithms, defining and learning uploaded shapes and objects. Now, we can identify an object even if we can only see 10% or 20% of it. The technology just leap-frogged, and I think we are now at a completely different stage. It has taken autonomous driving to a completely new level.

HDT: Let's talk about the fleets and your ultimate consumers and customers here. There are basically two different fleet business models for autonomous trucks. There’s the truck as a service model. And then there's the fleet that would actually own autonomous vehicles, their own fleet of driverless trucks. How does a fleet decide which way to go, or how to embrace this technology, whether to buy their own or contract another company as a service?

Buttler: I want to share [Daimler’s] core belief and also how we want to launch this technology.

It’s not our intention get into the business of our customers, so to speak, We have a clear agreement with Torc and also our other partner, [Waymo] that ultimately our vision is that the fleets who are today operating trucks, who are today providing logistics services, will also do that in the future. We want to provide a better means for them to do their job even more efficiently and safely. And that is, in this case, with an autonomous driving vehicle and the driver as a service.

Schmidt: First of all, logistics is much more than just driving. A lot of things need to happen to move goods safely from A to B, reliably, and at scale. And the fleets are just fantastic at that. Our goal is to enable them to do their business better, to deliver goods faster, safer, maybe cheaper. That should benefit the fleet and the end users. This is what we want to achieve, instead of competing with fleets and rebuilding what they’re already doing perfectly.

HDT: Is it a big step, do you think, for a fleet to take on an autonomous truck? How easy or difficult will it be to integrate this technology to fleet operations?

Schmidt: I think it will be a challenge, to be honest. That's why we are getting our customers involved early in the development process.

We have an advisory council consisting of key industry players from different segments of the industry, from parcel delivery to over-the-road fleets, to help us develop a product that is helpful for them.

As for the different aspects, it starts with simple operation and maintenance. You know, there's already a shortage of drivers, and also a shortage of technicians. That will have implications for both operations and maintenance in the future.

Then there’s the question of how to operate a mixed fleet. Let's be honest, it won't be all autonomous. So how do you operate a mixed fleet of manual-driven trucks and autonomous-driven trucks? It’s crucial that our trucks can seamlessly integrate into today’s mission controls. We are working hard to seamlessly integrate into customer processes and systems to make the switch and to use this technology and to make the first step an easy one.

HDT: The development cycle for autonomous trucks has gone on for longer than many people thought it would. We have seen a number of companies enter and exit the market already. Where do you think we're going to be in five years? Will we still dozens of autonomous vehicle companies out there, or just a few?

Buttler: We're all watching and observing what's happening in the market right now. We always believed that we would see some sort of consolidation in the market, and I think it’s starting to happen right now.

If we think five or 10 years ahead, we could see like three to four players surviving this game.

I think to be a player in the future, companies will need to have a couple of things in place.

One is a solid OEM partnership. I say that, very selfishly, being one of the providers, but we hear and see that in general, that this is definitely one of the success factors.

And the other success factor is to just stay focused from the autonomy side and be very, very good at the use case you decide to pursue.

HDT: A survey by the American Automobile Association say 68% pf those surveyed are afraid of driverless vehicles. And we know people are already afraid of trucks. How do you think motorists feel about autonomous trucks?

Buttler: It’s really important to build trust in the technology, by not just talking about how the industry is developing safety, but demonstrating it. The public will always see a driver behind the wheel and a safety engineer on the passenger seat when we test on public roads. This is to ensure that while we develop the technology, we have good oversight of the system. We will not take the driver out until we’re certain that the systems and the software can handle any driving situation safely without any human intervention.

Schmidt: Motorists will find our truck is really polite. It will never speed, it won’t cut you off, and it never gets aggressive. It has a really nice driving style, the kind you would like to see from all road users.