As we think about future trucks, the Messy Middle, and alternative powertrains, we also need to think about autonomous vehicles (AVs). While AVs are very much in the early stages of development, and it will likely take years (if not decades) before they are affordable and reliable, they offer the potential to improve safety and reduce operational expenses for fleets. Since they can save fuel, we can say AVs are a lever to help us decarbonize.
Over the last few years, there has been significant AV development. In the passenger car space, they are running fully autonomous (Level 4) vehicles in some cities. Tesla recently announced that one of its cars came off the production line and drove itself to the customer’s home, a 30-minute drive. In the commercial vehicle space, trucks are running in the southwest region of the country. This area has better, more predictable weather, and less congested roads than other regions of the country. Recently, Paccar and Aurora ran a truck along I-45 without a person in it. They later decided to operate the vehicle with an “observer” because we are still in the very early stages of AV development. Aurora just announced it will be running driverless at night on a route from Dallas to Houston. Kodiak is running with no humans on board on some private roads.
Dean Bushley, NACFE’s director of programs, has written several blogs that do a great job of summarizing AVs. Dean covers a variety of topics ranging from the benefits and challenges of AVs to the major AV players to the best use cases for AVs to comparisons between China and the U.S. AV leadership.
The benefits
As we think about the potential benefits of AVs, improved safety and reduced operational expenses come to mind.
Safety
The public and the trucking industry have major questions about whether AVs are safe. Can the engineers think about every potential scenario? Can the controls logic be designed to think equal to or better than a human? While there are great hopes that Artificial Intelligence (AI) will solve many of our future problems and be leveraged to successfully develop AVs, many people are skeptical as to whether all of this will really work.
I developed a different perspective while attending the National Private Truck Council conference in Orlando this past May. A presenter talked about AV safety. While I did not walk away concluding that AVs are perfect, the presentation opened my eyes to the fact that everything in life is relative. Yes, AVs will have some failures and accidents; however, it is worth comparing the track record of AVs to humans. Figure 1 shows that the experience of RoboTaxi is better than that of a human driver. The statistics show that the AVs have 80+% fewer accidents that result in airbag deployment and injuries.
Presentation at NPTC, given by Lawrence Bader
To a certain extent, this should not be a surprise. AVs have numerous cameras, LiDAR, radar, and many other monitoring and computing functions. In many cases, the sensing is far better than that of a human being. Aurora has a technology called FirstLight that can detect objects in the dark that are almost 1,500 feet away. That is nearly the distance of five football fields and far better than the human eye can see. It is advertised as being able to do this in the dark and can identify and differentiate among humans, vehicles, and debris.
In addition to this, it is worth thinking about the “dreadful D’s.” More specifically, AVs don’t have the flaws of humans. Think about all the drivers on the road who are experiencing the “dreadful D’s” — drugged, distracted, drowsy, disgruntled (road rage), and disoriented — and many more.
Operational costs
ATRI recently updated its annual study of the costs of trucking, and it shows the two largest costs of operating a truck are driver wages/benefits and fuel. The cost of driver wages and benefits is approximately $1/mile while the cost of fuel is about $0.48/mile. Combining these two accounts for about 65% of the total cost of operating a truck.
We are still at the early stages of AV development, but if we can successfully implement Level 4 autonomy, the driver can be eliminated, saving a large cost.
As for fuel savings, Aurora issued a paper which estimates (based on simulation) how fuel can be saved. Many of these concepts can theoretically be done with modern-day controls alone without AV technology, but because of the compensation structure for drivers and the discipline and precision of a computer, the AV provides an advantage.
Take a look at some of them:
- Net energy efficiency: AVs can provide a 13% to 32% net energy efficiency improvement per loaded mile. This translates directly to reduced fuel consumption and is a contributor to decarbonization.
- Lower vehicle speeds: Most drivers are compensated on a dollar/mile basis. There is a natural tendency to run at higher speeds, especially under ideal driving conditions. With an AV, you can easily set the cruise control back to 60 or 65 mph and save a significant amount of fuel. Thinking of my summer family vacation this year driving through Texas, many of the speed limits were 75 mph with traffic flowing at 80 mph. If an AV can run “slow and steady” at 60 mph, a great amount of fuel can be saved.
- No idle: Since there is no driver, there is no need to idle an engine or run the HVAC. This can translate to significant savings.
- Minimizes Deadheading: It is believed that AVs will be able to minimize deadheading because there is no driver and therefore no need to return the truck in order to get the driver back home. An AV won’t deliver this by itself. Route and network optimization will be needed.
- ECO drive: The truck can be driven in the economy mode (ECO), meaning it will slowly accelerate, decelerate, etc., maximizing the conservation of energy and thereby saving fuel.
- Off-peak operation: Since AVs can run nearly 24 hours a day, a good portion of their time on the road will be at night and during off-peak traffic congestion periods, which saves fuel.
- Vehicle optimization: If we look at the Tesla Semi or Windrose electric vehicles, we can see the central drive and design from a clean sheet of paper. These trucks have been designed to optimize freight efficiency. The same can be said for an AV. Think of cab requirements needed for the driver so there is good visibility and comfort. Think of sleeper cabs and the extra weight and design constraints they impose. The truck of the future can start from a clean sheet of paper and optimize the Coefficient of Drag.
New products and innovation arise
As a kid, I remember debates as to whether going to the moon made any sense. I can remember my two uncles debating this topic with some passion over a couple of beers. But a lot of great things came out of our investment in that journey. Examples include Computed Tomography (CAT) scans, camera phones, cordless tools, memory foam, and freeze-dried food (well, the last one might not be a compelling example!).
The development of AVs also is bringing new technology. When we think of AVs, we think of the end goal, which is driverless vehicles. However, today a lot of our cars and trucks are using many of the stepping stones of AVs. First and foremost are the Advanced Driver Assistance Systems (ADAS) features such as lane departure warning, blind spot monitoring, driver wheel torque control, dark assist, etc. AV development also relies heavily on AI, high computational capability, and advanced mapping. AVs are contributing to the development of this innovation.
Challenges
While AVs have many benefits, they are not without their challenges. Some of the biggest challenges include the following.
Costs: Cost is the biggest challenge. Companies such as Aurora, Kodiak and others are investing billions of dollars in this technology. None have made a profit to date. If we want to look at the costs at a vehicle level, Figure 2 is a nice illustration. For the sake of this discussion, we will focus on the U.S. trucks and compare Level 0 and Level 4 automation. This basically compares a truck with no autonomy (Level 0) to a full driverless vehicle (Level 4). You can see the total costs of a U.S. level 4 truck is approximately $250,000 less than that of the U.S. Level 0 truck. The significant savings is the result of a reduction of about $400,000 in driver wages. There also are some fuel savings. The upfront cost of a Level 4 vehicle is approximately $150,000 more than that of the Level 0 vehicle. The key will be to maximize truck utilization. This is similar to comparing EVs to internal combustion engines (ICE) in that there is a significant upfront cost for the EV, but the cost of operation is significantly lower. In the case of an EV, the key is to utilize it a lot and use cheaper and more efficient energy (electricity) instead of diesel fuel. In the case of the AV, the same idea holds true in that it is better to highly utilize it, thus generating more revenue in a 24-hour period and thus saving on driver pay.
Edge AI Vision- EV compatibility (or utilization): This is a topic that I find a little unclear. Some people will say that there is great compatibility between an EV and an AV. It is true that there is good compatibility from an electrical architecture standpoint (e.g., advanced sensors, data handling, controls). However, because of the high costs of AVs, high utilization is essential. The truck needs to be on the road hauling freight and making money to offset its high upfront costs. When comparing an EV to a diesel truck, the EV will take more time to “refuel” even when considering fast charging. The time to recharge the batteries is time not on the road hauling freight. So, while there might be some unique cases, the use case for AVs will be better for ICE vehicles than for EVs. As we think of the future and hopefully cleaner vehicles, I believe EVs will be at a disadvantage as AVs. Range and fast charging need significant improvement to build the business case for an electric autonomous vehicle.
- Performance: Can the manufacturers of AVs get them to work and perform equal to or better than a human? Will they really work? Will they be reliable? These are the $64,000 questions — and the verdict is still out.
However, today’s technology and sensing capability of cameras, LiDAR, radar, etc. can identify objects better than the human eye. The designers are creating high-fidelity simulation scenarios that test millions of different combinations. Via telematics, data on millions of miles is being captured, gathering details of different roads, terrain, buildings, etc. There’s also redundant processing of different sensors. The software logic is designed to read and compare all these inputs. Data analytics and AI are used in this process.
Finally, all of this is continually being updated and revised as errors occur. As items are missed, they are noted and corrected. Yes, there will be misses and errors, but given AI and computational capabilities, new scenarios and combinations will be added.
We have a long way to go to see a large number of autonomous trucks traveling down the local interstate on sunny days, let alone on days and nights with rain, wind, and snow. Also, the upfront costs of an AV are extremely high, and given the tight margins that fleets operate on, this is a significant challenge. However, we need to think over at least a 10-year period and realize that innovation will prevail, and these AVs will create a noteworthy presence in the commercial vehicle space.
There is a good chance we will see limited Level 4 corridors in five to seven years and broader adoption in 10 to 15 years. This will result in lower operational expenses and a reduction in fuel or energy use.
AVs will be a lever to help us reduce emissions and decarbonize. They will be a legitimate addition to the Messy Middle.
