Aug. 2, 2021
Autonomous trucking will change the trucking industry more over the coming decade than it has during the past five decades.
The trucking industry has been more or less the same over the past half-century, little changed by major technology advancements we have witnessed in other core parts of the economy such as medicine and information technology. Autonomous trucking (also known as "self-driving trucks") is emerging as the most disruptive change in our lifetime, and it will transform the industry as we know it today.
Self-driving trucks are similar to self-driving cars in the sense that an onboard computer acts as the driver. This technology is relatively new, and it is enabled by recent advancements in artificial intelligence, computer vision and hardware improvements.
Computers historically have been very good at defined tasks. Computers were originally designed to take a set of instructions ("computer code", such as do task 1, then task 2, etc) and follow those instructions step-by-step. Computers are able to process code extremely fast and more efficiently than humans; however, historically, computers were limited to only following those instructions and not "thinking" like humans.
Previously the only way you could "drive" a truck with computer code was by telling the computer exactly what to do. For instance, accelerate to 30 miles per hour, then turn right after 30 seconds. This code can work perfectly in a controlled environment without traffic and pedestrians, but it fails in real-life applications. It is impossible to have explicit instructions for all roads and all environments.
Artificial intelligence changes the way computers work - it enables computers to make decisions and perform tasks without receiving explicit instructions. Advancements in neural networks and computer hardware are the key components that enabled this change. Now you can show a computer video recorded from the dashboard it can determine that there is a stop sign in one location, a pedestrian walking in another location, and a car approaching from the left side, and then decide how to steer the vehicle to make sure everyone is safe and the vehicle gets to its target location (similar to how our brain makes sense of what we see when we drive).
Artificial intelligence is created by programmers "teaching" computers how something works. This involves showing the computer a lot of data (for instance, millions of hours of videos with trucks driving on the road), and over time the computer learns from these data (similar to how your brain neurons change when you learn a new task). After a lot of training and time, the computer becomes "intelligent" with that given task because it has seen that task performed many other times and learned what works and what does not work.
Artificial intelligence in self-driving cars and trucks collects data input from cameras, LIDAR, and other sensors, similar to how our brain collects information from what we see, hear and smell. It then interprets the data (for instance, "that looks like a red stop sign") to understand the environment, then it makes a decision of what to do in that environment ("it's a stop sign, I should start braking").
Computer vision powered by YOLO
There is a common fear among some drivers today that autonomous driving will take their jobs. The truth is, nobody knows for sure, but many industry experts believe it is decades away at the earliest (if at all).
Steve Nadig, Director - Mechatronics-Power Train, Body Builder and Chassis Systems at Daimler Trucks North America, said in a recent interview, "Daimler Trucks is going to take it step-by-step, safety-by-safety, use case by use case, to make sure that we are going to put the safest truck on the road possible... I think in 10 years we will still have a driver in the seat. At this point, I have not seen the evidence to take the driver out of the seat."
Today there are over 2 million truck drivers in the United States who keep goods moving from ports to homes, and factories to job sites. Trucking is a critical industry and necessary to ensure commerce flows. Trucking will not go away any time soon.
The closest case study of what could happen with the trucking industry is what happened with the airline industry. Today computers fly most commercial aircraft, and humans are only needed for takeoff, landing and special situations such as strong turbulence. The rest of the time, the pilots can kick back and enjoy the blue skies as computers navigate the clouds and ensure the safety of hundreds of human lives.
Trucking will likely follow a similar path as aircraft. A driver will be needed to ensure safety (for instance, to takeover if road conditions become challenging), attach loads, tend to the tractor or trailer if something fails, etc. Computers will assume more responsibility for driving on long highway stretches under normal driving conditions.
Similar to aircraft, the cost of failure in trucking is serious and impacts lives. As with any technology, it takes time to perfect it, and the initial versions have bugs that need to be worked out. The same will go with autonomous driving. Automated driving needs to be 99.9999% safe, not just 99%.
Automated driving may give more leeway to change Hours of Service regulations because computers will enable to truck drivers to go longer distances and work more hours with less time with "hands on the wheel."
Similar to how there are many shades of gray, there are five levels of computer-assisted driving. Currently, advanced autonomous driving systems in cars are Level 3 (Conditional Automation where a human driver is still required).
Whether you know it or not, you have probably already interacted with autonomous driving. Adaptive cruise control and lane assist are both forms of computer-controlled driving.
Image courtesy of Synopsys
Cars are expected to reach Level 5 automation (full automation) in the next 5 to 10 years. Trucking may take even longer, given trucks are more complex to operate (they are heavier and larger).
Similar to the trend we saw in automobiles where there were a few early leaders and then the whole industry moved to follow, the entire trucking industry will likely develop semi-autonomous trucks down the road.
When an industry technology changes, it invites new market entrants (such as Tesla, etc), while some market players become prone to dying off. It is the natural innovation evolution cycle.
Companies currently developing semi-autonomous trucks include (as of today, more are launching):
Companies will continue investing billions of dollars into self-driving technology, and we will certainly hand over more control to computers in the future. Like many other things, this change will take time as the technology is perfected.
Trucking is a critical industry that keeps our economy flowing. It also requires a high level of attention to safety. Given these factors, it is likely industry participants and regulators will want to ease into this new era gradually.