Cruising down the highway, it seems as if everything is just as it should be: visibility is good and traffic is moving well. Suddenly an improperly secured mattress flies off the roof of the sport utility in front of you. There is no way to stop in time and no way to avoid the flying bed. Because you’re aware of your surroundings, you know that immediately to your right in the next lane over is a bus full of school children. To your left is a mini-van full of seniors. In a split second, you’ve got to decide what course of action you’re going to take. Do you swerve left and hit the van, cut hard right and collide with the bus, or keep driving straight on, hoping that the impact with the mattress won’t make you lose complete control?
Actually, you do none of the above. Instead, your autonomous, self-driving truck makes the decision for you, but just what decision it makes will depend on how its software engineers have programmed the truck’s algorithms to minimize harm. In some cases the result might be to take the action that is predicted to injure (or kill) the smallest number of people. In others, they might have taught the truck to crash into the larger vehicle, assuming it can take the brunt of the hit better. Programmers may have even given the vehicle a preference to take every action possible to avoid injuring children, or to do everything in its power to ensure that you, the driver, survives unscathed.
“For the trucking industry, the risk is magnified: we’re no longer talking about two-tonne vehicles of steel and glass, but up to 40-tonne vehicles that may be carrying hazardous freight. So it becomes even more important that manufacturers get crash-optimization right with trucks, given the risk of much more serious—and high profile—harm,” said Patrick Lin, director of the ethics and emerging sciences group and associate philosophy professor at California Polytechnic State University in San Luis, California.
While this may seem like a fantastic scenario, it’s exactly the type of problem artificial intelligence researchers are facing in their quest to produce self-driving vehicles. It’s also the ethical issue that confronts those like Lin whose jobs require them to think about the implications of how society is going to incorporate the latest technological developments.
Whether we are ready for it or not, the technology that will enable vehicles to propel themselves down the road with minimal or no input from a driver is either already here or rapidly approaching. Responses from governments, regulatory agencies and other interested bodies addressing the use of autonomous technologies are still forthcoming, however, especially when trucks are involved.
Seeing is believing
Recent months have seen a flurry of demonstrations and announcements from manufacturers anxious to demonstrate the latest developments in their autonomous or driver-assisted technology. ZF, for exampled, showed off a truck piloted by remote control from a tablet. Peterbilt’s truck navigated a course using GPS points. Daimler first demonstrated how its autonomous truck could drive amongst other vehicles on the highway and then, on a later date, deliver a senior company executive to a press conference inside an airplane hanger. Peloton sent a platoon convoy down the I-75 in Michigan. While none of these capabilities are currently offered on commercial vehicles, the time when they will be is getting ever closer.
John MacRitchie started out as a bit of a sceptic when it came to autonomous vehicles, but after investigating the topic, he believes we are on the verge of seeing self-driving vehicles on the road. MacRitchie is the regional director of the Ontario Centres of Excellence (COE), a not-for-profit organization funded by the provincial government to help Ontario-based businesses, universities and research organizations bring innovative technology to market quicker. Currently, through the connected vehicle/autonomous vehicle research program (CVAV), COE is operating a pilot project and distributing $950,000 to companies working on self-driving technologies for cars, buses or commercial trucks.
“We’re technically a lot closer to having that [autonomous] capability than people realize,” said MacRitchie.
“When I began to look at this area, the thing that surprised me was how quickly some of the companies that are in this space felt this would happen—2020 is not out of the question. It’s that soon. From an automotive standpoint, that’s a very short period of time. That’s only about a cycle-and-a-half or maybe two in terms of vehicles between now and then.”
Given that it’s part of MacRitchie’s job description to be upbeat about new technology, some people might be willing to dismiss his timeline projection, but even vehicle manufacturers are talking in terms of years, not decades, and are saying much of the hardware is already in place on their trucks.
“The big picture is two or three years out for lane-keeping, parking assist, traffic assist and technologies like that. Then manned vehicle platooning,” said Bill Kahn, Peterbilt principal engineer and manager of advanced concepts. “Then we step out a little bit farther and we give the truck the ability to manoeuvre around items. Stepping out a bit farther, that’s when we get the possibility of the truck being able to drive urban duty cycles with input from the driver.”
A Peterbilt forgoes driver input and uses GPS to navigate around an obstacle course.
According to Kahn, Peterbilts already come with many of the pieces in place to make the trucks more autonomous.
“They would require an upgraded camera system and an upgraded radar system to what we have today, but that’s something that is already on the product plan for all our trucks. The steering column we are using now is already seeing some use [in autonomous testing] because of its ability to self-centre. If you’re on a crowded road or get a crosswind, it automatically re-centres the steering wheel to zero so it reduces the fatigue on the driver. That’s an option that can be done next year. The computer controller that ties it all together is not that hard. Basically, we developed this autopilot system on this truck in about two months once we knew we had control of the vehicle.”
Volvo Trucks also has plans to install much of the foundational technology on upcoming models, and Carl Johan Almqvist, the company’s traffic and products safety director, said adding more advanced driver-assist or autonomous technology may “not necessarily” result in pricier trucks.
“Pretty much most of the technology is on the way into the trucks already. In Europe, by the first of November next year, all trucks will have to have collision warnings in them and emergency braking systems. Then you’ve got the radar. The brakes are in place. We are already equipping a lot of our European trucks with Volvo Dynamic Steering, which is a steering aid. The only thing that is really missing is the communication between those components.”
At this point there are three separate classifications that are useful to keep in mind: connected vehicles, driver-assisted vehicles and autonomous vehicles, and manufacturers are very circumspect about what descriptions they use for their vehicles and products.
Connected vehicles are those which transmit data back and forth between other cars and trucks and can even “talk” to the infrastructure. Weather Telematics, an Ottawa-based company working with the COE, for example, places sensors—essentially miniature weather stations—onto truck roofs with the goal of collecting highly localized weather data and transmitting that information to other vehicles heading toward that area. Combining that type of information along with data gathered from a highway alert network, would allow drivers (or autonomous vehicles) to better understand the conditions they are travelling in and adjust their driving styles accordingly.
Driver-assisted vehicles and technologies are phrases used to describe either what is available now (blind spot warnings, for example) or what will be available in the very near future. Daimler describes these types of technologies as “freeing the driver from monotonous driving tasks, to provide support in highly complex driving situations, and in doing so to advance on the road toward accident-free driving.” According to Daimler, the transition from these technologies to the “long-term objective of autonomous driving will be gradual—hand-in-hand with an increasing level of automation of the assistance functions and the connectivity of vehicles with each other and the infrastructure.”
Currently, much of the connected and autonomous vehicle technologies are being developed independently.
“The connected vehicle technology is orthogonal to the autonomous vehicle, but at some point the two will merge and we will gain the benefits of both,” said Paul Godsmark, founder and CTO of the Canadian Automated Vehicles Centre of Excellence (CAVCOE), a consultancy group offering advice about preparing for a world containing self-driving vehicles.
Keeping data safe
Even though he is firmly convinced of the economic and social benefits of autonomous, driver assisted and connected vehicles, Godsmark understands that they do offer some challenges, and one of the major ones is around data management and security.
“People talk about there being a data revolution with connected vehicles, but with the autonomous vehicles we’ll have a magnitude more than with connected vehicles. We will be able to do so many good things with that data, even though it has so many privacy issues. That’s a whole new debate that needs to be taken up, but for some reason, it hasn’t been given the air time it needs,” he said.
Godsmark explained that with cameras capturing real-time video and LIDAR (laser-based remote sensing technology) systems tracking cars and trucks, we will have a “real-time, dynamic view of streets—how many vehicles are on the road, how many pedestrians, how many people sat outside at a café. We will have real time information that’s of huge value. Yes, it is incredibly creepy and scary and incredibly powerful and empowering.”
All of these systems won’t just keep track of what’s going on in the outside world, they will also keep a close eye on activities in the truck as well, said Lin.
“For the trucking industry, I’d expect that privacy issues will be a top concern. Truck drivers are already held to near-superhuman standards to make their trips in a dangerously short amount of time. Add to this the push for telematics by the insurance industry, driver-drowsiness detection and other biometric technologies that monitor humans inside the vehicle and the general appetite for interconnectivity and data collection by the technology industry, and it looks like the perfect storm for Big Brother inside the vehicle. The transportation industry may be affected first, since this would be done in the name of workplace safety.”
Even if the industry manages to figure out how to ethically and responsibly use all of the data it collects, without making the driver feel like he or she is under criminal surveillance, there are others who are less scrupulous with their intentions.
“My biggest concern with autonomous vehicles is the whole cyber security side,” said Godsmark. “A malicious actor could choose to take control of individual vehicles or a fleet of vehicles and do all sorts of damage. That does scare me terribly. But at the same time, those malicious actors can do pretty much the same thing with people in the vehicles or with unwitting people driving vehicles. The fact is although there is a very high severity associated with it, there is a very, very low risk.”
Problems with people
In tightly controlled situations, there is no reason why autonomous vehicles can’t function well. Some mining companies, including Suncor in Alberta are already using self-driving trucks at excavation and extraction sites. But when you add a large number humans into the mix, that’s a whole different story. People are unpredictable. They don’t always make logical decisions. Plus they make mistakes. So designers and programmers of autonomous vehicles have to build in enough intelligence to allow the trucks to react when the unexpected happens.
“That’s the big challenge: that we have to have the systems try to figure out what the human being is going to do, and the human being can do pretty much anything. Definitely the parts where we try to predict what happens with other traffic is a research area we will be looking into—trying to get a picture and understand what is going on around the vehicle,” said Almqvist.
It’s not just trucks that have to adapt to people. People also have to get used to the idea of trucks cruising on autopilot, especially if they are driving along a highway beside a lot of experimental, self-driving trucks, as Marlo Anderson hopes will soon happen. Anderson is the spokesperson for the Central North American Trade Corridor Association (CNATCA), a group of individuals and local governments attempting to turn Highway 83 (which runs north-south from Canada to Mexico through the middle of the US) into an autonomous trade corridor that will allow manufacturers and operators of self-driving cars and trucks to test and operate their vehicles. Besides the land vehicles, the association wants to create a 40-mile-wide fly zone for drones.
“I think the safety thing is going to be a big deal. We get that question a lot. Millennials adapt to technology pretty easily, but I’ve had this question brought up a few times: ‘What is my grandmother in rural South Dakota going to feel about being alongside this huge flatbed loaded with oil equipment and there is nobody driving it?’ I get that. I really do,” said Anderson. He added that people need to alter their perspectives when worrying about the safety aspects of road-going autonomous vehicles.
“When we fly, the entire flight is autonomous. The only reason the pilots are there is people need that feeling of safety, but it is actually the instruments that fly safer. When you think about it, would you rather trust instruments or pilots when flying through clouds? You don’t have to worry about pilot fatigue. For every argument you can make for a vehicle not being driven autonomously, there is an equal and opposite argument for a vehicle that is driving autonomously.”
Setting the law
CNATCA is attempting to convince US lawmakers to create legislation that will allow the use of self-driving and driver-assisted vehicles on the highway. Anderson described it as both a time-consuming and time-sensitive process.
“Over the next few months, we are starting to meet the entities in different states that will be involved. Of course we want to recruit the legislators in each of the states that we feel will be the key individuals to help us move forward. It’s a bit of a process because there are six states involved. It’s not just something the federal government can sign off on. A lot of the roads are state-owned roads. Highway 83 is basically state-regulated,” he said.
“We are working on the legislative packets and will be introducing those in the next legislative session, which is early 2015. That’s why we’re so aggressive to meet all of the legislators. Most of the states are in a two-year cycle, so if we don’t get this done in 2015, we’ll have to wait until 2017. At the minimum we’re hoping for some limited legislation. If it plays out well, we would have a corridor that would run 1,800 miles by the middle of 2015, but I highly doubt that will happen. Maybe we’ll have something between Bismark, North Dakota and Pierre, South Dakota where we’ll have sections of roads where we can start proving the technologies.”
Even if CNATCA only manages to get a small portion of the highway approved for autonomous vehicle testing, the state (or states) involved will still be miles ahead of Canada. So will Nevada, Florida and California as they have already made provisions for the testing of these emerging technologies.
The ZF Innovation truck is remotely backed through a traffic cones.
On the federal level, Transport Canada is doing a bit of investigative work, but is taking no steps beyond that. According to Transport Canada spokesperson Karine Martel, one technology that Transport Canada is monitoring is co-operative truck platooning systems, or CTPS. CTPS use specialized software and communications technologies to semi-automate highway driving and platooning of heavy-duty trucks.
“Transport Canada is also testing autonomous emergency braking systems in passenger cars with a view to determining if regulated performance requirements for such systems may be needed. These systems are also available on a voluntarily fitted basis in Canada now, on some new commercial trucks.”
It seems as if the government is leaving the door open for manufacturers who want to bring highly automated vehicles into the country. “Transport Canada would permit the importation of an automated vehicle, provided that it has been certified by the manufacturer as complying with the current safety standards that apply to the vehicle class,” said Martel.
As mentioned earlier, Suncor Energy is running a pilot project in Fort McMurray to test what it calls its autonomous haulage system (AHS), but because the vehicle (reportedly from Komatsu) doesn’t operate on public roads, the province has no need or authority to regulate its use. And even though awareness has been raised about the capabilities of autonomous technology, Alberta is not even considering making legislative changes to allow self-driving vehicles on provincial streets or highways.
Ontario by comparison is miles ahead in the autonomous vehicle race, even if it is still in the starting gate.
“As part of our monitoring, on December 17, 2013, the Ontario ministry of transportation posted a potential autonomous vehicle pilot framework on the Ontario Regulatory Registry for consultation,” explained spokesperson Ajay Woozageer. “The potential pilot looked to safely test and evaluate fully autonomous vehicles under prescribed conditions before they become widely available to the public. The posting was open to the public for 45 days and closed on February 24, 2014.”
The proposal called for the pilot project to run for five years, and apply to passenger cars and light trucks only. Since the comment closing date, however, there has been no word about the status of the project. CAVCOE’s Godsmark believes the “project got shelved during the provincial election. Now we are waiting to hear what Ontario will do to move forward.”
One of the reasons governments may be reluctant to jump on the self-driving bandwagon is that right now there are no good answers to the question ‘who is responsible if an autonomous vehicle gets into an accident?’ If there is no driver to blame for not paying attention or failing to make a proper lane change, does the liability fall to the fleet operating the truck, or the vehicle manufacturer who built the truck? What about the third-party OEM that provided the logic controllers or who wrote the collision-avoidance code that sent the truck careening into that bus full of school children? Governments themselves may come under extra scrutiny, said Godsmark. “You’ve got the road authority, as the road layout and engineering becomes absolutely key now… Driver error is the cause or major factor in 90% of crashes. What will happen with autonomous vehicles is as we mostly remove the driver-error component we’ll discover deficiencies in the road are actually contributing to crashes,” he said.
“The finger will be pointed at every single one of those and the law will find out where the liability lies. Once that happens, we set a precedent and move forward,” he said adding that he expects there to be early, precedent-setting lawsuits for billions of dollars, but after the first few are settled, it will start to become obvious what the law of the land thinks about who holds responsibility and liability.
Godsmark also said that he could see manufacturers and equipment makers combining funds into a common insurance pool designed to handle the initial accident claims before the insurance companies become comfortable adding self-driving trucks to company policies.
Creating the support system
Another reason governments may be unwilling to encourage autonomous vehicles are the extra demands that may be placed on them to improve infrastructure or change procedures.
“You see a lot of discussion about V2I which is vehicle-to-infrastructure communications,” said Kahn “and which require a government or a municipality to install [smart] infrastructure and then maintain it. It’s just something we at Peterbilt don’t want to focus on if we have the ability to have the vehicle be situationally aware of its surroundings. That’s why GPS navigation is so important to us.”
An in-traffic demonstration of platooning by Peloton Technology.
As well, governments will have to devise and oversee new tests and inspection procedures to certify that the self-driving trucks are safe enough to be on the road, all of which will cost time and money, although Godsmark predicts that any costs associated with turning our highways over to self-driving vehicles will be more than off-set by the monetary benefits they will provide. He cites figures that say road crashes cost Canada $62 billion per year, or 4.9% of our GDP. In contrast, he says investing in autonomous technology returns between 10% and 20% in cost-savings.
“Owners and operators will potentially make a lot of money while drivers can potentially lose out big time. Right or wrong, overall society will benefit from lower costs of goods in the shops, but at what price for the truckers’ job?” he asked. He added just because there is a driver shortage now, it doesn’t mean that situation will continue once self-driving technology proves itself. Then drivers will be reduced to chaperones, will be brought inside to handle more administrative duties, or will have to be retrained for other jobs. “I think we should be preparing now and having plans in place so when it does happen we will have a good idea what to do.”
Even for those who are sceptical that self-driving trucks will soon be popular enough to put drivers out of work, it’s still important that society as a whole gives consideration to the emergence of autonomous technology, so we are ready to deal with it when it finally does arrive—even if it winds up presenting itself in a different form that originally predicted.
“Now the technology has come so far, that’s when you start thinking ‘is this really what we want to do? Why do we want to do it?” said Almqvist
<h2>On the road today</h2>
Recently manufacturers have been demonstrating a their latest accomplishments in autonomous vehicle technology.
Daimler self-driver: the truck of the future
Remote control ZF: steering by tablet
GPS-guided Peterbilt: hands-off driving
Peloton Platoon: connected traveling down the 1-75