The Institute of Road Transport Engineers of New Zealand held its biennial conference in Rotorua in August. The theme this year was Technology and Infrastructure, arguably the two most pertinent topics of the day. As a result, we‘ll feature extended coverage of the conference over the next few months in order to bring to you commentary on these important subjects from leading academics and industry minds.

There‘s an underlying pessimism among the global trucking fraternity that the rise of vehicle automation will lead to the redundancy of the driver. And rightly so; as automated vehicle technology becomes ever more pervasive, the core skill sets of the truck driver will need to evolve with each successive model generation – to the point where the driver‘s job will be far removed from how it‘s seen today. However, John Woodrooffe, principal of Woodrooffe Dynamics (a consultancy focused on commercial vehicle policy, safety, productivity, energy use and emissions) doesn‘t think we‘ll see fully autonomous vehicles on the roads any time soon. While there‘s great enthusiasm for driverless vehicles – with levels one and two achieved and implemented in many modern vehicles – there are significant challenges facing them too, he says.

When it comes to levels three and four, where the driver isn‘t completely involved and the vehicle controls all aspects of the driving task under certain conditions, Woodrooffe suggests that there is a higher risk of high-speed crashes should the driver need to resume control. “We‘ve seen some significant crashes with automated vehicles simply because the driver tends to disassociate himself with the vehicle state.” Level five – full automation where the vehicle wouldn‘t even need a steering wheel – may be unobtainable in the foreseeable future, says Woodrooffe. “Imagining how this will happen is really hard for me.” This is because for vehicle automation to be truly effective, says Woodrooffe, it will need to surpass our own rate of crash avoidance. “Since most crashes are the result of human error, we think automated vehicles should solve that issue. But crashes are rare events – as clumsy, awkward and messy as a human is, we are pretty successful at avoiding them.

Photo: John Woodrooffe

The chance of a human controlled vehicle being fatality-free during a given year is 99.99 percent. “For any machine system to have that sort of reliability is a tall ask. One of the greatest challenges for high-level selfdriving vehicles will be to achieve the current levels of human crash avoidance performance, to the same level of success.” So what‘s stopping the artificial machine from beating the organic being? The first issue is a question of nuance. “Artificial intelligence [which forms the basis of self-driving tech] is the acquisition of information and rules to reach approximate or defined conclusions with the capability of self-correction. It‘s more predictable, robotic in nature and behaviour,” explains Woodrooffe. “It handles well in terms of the normal and predictable but the surprises, the unintended, are challenging for it to decipher.

“When thinking of human intelligence, it‘s the intellectual prowess of humans that‘s marked by complex cognitive feats and high levels of motivation and selfawareness; well-defined nuanced behaviour.” This brings into question the machine‘s ability to operate unaffected in everyday conditions marred by erratic road users of all kinds, unpredictable weather conditions, and changing road conditions (Woodrooffe references New Zealand‘s complex geography and tight roads). The vehicle, itself, will also age over time and sit at the mercy of humans adhering to (potentially expensive) maintenance schedules.

“We may be faced with dialling back our expectations,” he says. “We‘ve seen the demonstration of fully autonomous vehicles under ideal conditions and limited mileage. We‘ve also seen partially autonomous vehicles introduced in larger numbers and being involved in crashes. Our present automated vehicles have relatively poor intelligence based primarily on sensor input and software situation analysis,” Woodrooffe says. He suggests that with engineering comes a certain arrogance and says this was demonstrated by the recent Boeing 737 MAX disasters. “When we unpack those disasters, flawed artificial intelligence was a contributing factor and that‘s a window to the challenge we have to bring to the automotive side.

There was a lack of human intelligence applied during the system design and since AI is the simulation of human intelligence, we have much to be concerned about.” So, for now, the human element – the driver – is safe. And Woodrooffe says that, over time, there will be a greater appreciation for what humans can do when it comes to controlling vehicles. “I think we‘ll also develop a strategy where, instead of replacing drivers, this technology will fortify them.” That is to say, when the vehicle experiences critical conditions that lead to crashes it can then intervene for the human – instead of it being the other way around. “We want to see vehicles that refuse to crash and we‘d love to see this development now put an emphasis on driver support. As that technology matures and as it shows that it can handle the extraordinary events, then we will be at a point to move to full autonomy.” Good news, then, for the world‘s truck drivers.