SINGAPORE (July 24): On an overcast summer’s day, in the outskirts of Gothenburg, the port city on the southwestern coast of Sweden, three trailer trucks are lined up along a wide, oval track. On signal, they start off in unison, one close behind the other. After a while, the drivers in the second and third trucks pick up their radios to talk to each other, gesturing; their hands are completely off the steering wheel.
These trucks, equipped with autonomous driving technology, are part of vehicle manufacturer Volvo’s trials that have been running since September 2016. The company, best known for its focus on safety and the hardy construction of its vehicles, has had its trucks put to the test in various sectors, including mining, agriculture and even in garbage collection. A self-driving truck was deployed around the clock in an underground mineral mine in Kristineberg, north of Stockholm. The company has also partnered with an industrial farmer in Brazil, piloting an autonomous truck for sugarcane harvesting that cuts crop damage.
Importantly, the trucks have so far been deployed only in industrial or commercial, and completely controlled, environments. And, for Volvo, there are still numerous variables to figure out — inclement weather, for instance.
“The [necessary technology and] components are there. When you connect them together, you have an autonomous truck,” says Lars Erik Forsbergh, managing director for Volvo Trucks in Southeast Asia and Japan. “The challenge is, how do you integrate that into the traffic environment in a safe way?”
Hayder Wokil, Volvo Trucks’ mobility and automation director, explains: “The [technology] is not mature enough to handle all weather conditions. [For example,] when it rains heavily, the radar in my car, a 2017 Volvo, stops working. When I have the [adaptive cruise control] on, it’ll tell me there’s a malfunction. How will it work with autonomous vehicles? The technology still needs a lot of time to develop to handle all public situations.”
Just a few years ago, self-driving cars seemed to have arrived. In Singapore, transport authorities and planners, in particular, were keen on developing the technology for use in the public transport system. It seemed ideal: Singapore’s small yet efficiently designed infrastructure provided the perfect backdrop for a system that promises to alleviate traffic and manpower woes.
“It is very important to test the vehicles in scenarios and areas that will teach your software routine [and] facts that are generalisable,” says Walter Theseira, professor of economics at the Singapore University of Social Sciences (SUSS), and a frequent commentator on Singapore’s transport system. “Singapore meets that criteria because the road network is fairly standard in international terms, and I think it is sufficiently challenging that your software will learn how to deal with different road users and pedestrians.”
Trials were set up, and in January 2015, the research hub of One-North was designated as a test-bed for self-driving vehicles. In October 2015, Prime Minister Lee Hsien Loong was ferried in an A*STAR self-driving vehicle. Other pilots were being run for autonomous buses within a university campus, and shuttle pods within Gardens by the Bay.
Meanwhile, ride-hailing company Uber was conducting driverless car trials in Arizona and Pittsburgh in the US. In 2014, electric car maker Tesla offered an Autopilot feature in its Model S cars.
But fatal accidents on March 18 with Uber’s driverless car trials in Tempe, Arizona and on March 23 with Tesla’s autopilot system in Mountain View, California halted those ambitions. In Singapore, start-up Nutonomy had been conducting pilots at One-North when its driverless car crashed into a pickup truck in October 2016. The company has since been acquired by Delphi Technologies, and its trials have resumed.
Those accidents occurred when the vehicle’s radar failed to recognise an obstacle or the vehicle failed to stop. Presumably, it is back to the drawing board for the engineers. What has become of those trials, particularly in Singapore? How far away are we from self-driving reality? What will it take to make it happen?
Ready, set, no?
Singapore has consistently ranked among the top three countries, globally, and the top in Asia in terms of autonomous vehicle readiness. According to the 2018 KPMG Autonomous Vehicles Readiness Index, this is due to the country’s policies, infrastructure and consumer acceptance.
The only shortcoming that held Singapore back, in the infrastructure category, was the lack of electric charging stations, according to Satya Ramamurthy, head of infrastructure, government and healthcare at KPMG. “From a technology and innovation point of view, Singapore has done a lot in moving the needle on technology,” says Ramamurthy.
The move to autonomous vehicles has been pegged as part of the larger Smart Nation initiative. The government has partnered with private organisations and research institutes to run trials over the past few years. These include an electric bus trial run by the Land Transport Authority and ST Engineering Land Systems, a unit of Singapore Technologies Engineering; and a collaboration between the Ministry of Transport and PSA with automotive manufacturers Scania and Toyota Motor to develop technologies for autonomous heavy vehicles.
But while the use of autonomous vehicles in the industrial sector continues to make progress, authorities seem to be still grappling with self-driving vehicles for consumers. The technology development also seems to be skewed towards public purposes, notes Ramamurthy.
He adds that, in Singapore, the deployment of autonomous vehicles would most likely start in so-called “closed loop” environments, “in places such as Sentosa, where sensors can be fitted to guide the vehicle before open-loop autonomous vehicles can come into play. A lot of data that can be collected will improve the safety standards and [facilitate] the introduction of the open-loop autonomous vehicle as well”.
When can autonomous vehicles be safely deployed on open roads? “The reality of this, as with any such innovation, is two steps forward, one step back. Whenever there is an accident, people will wonder: Are we doing the right thing?” says Ramamurthy. “Singapore is as good a place as any to test for safety. In fact, it has some special characteristics such as sudden, heavy tropical rainfall [which makes it good to test for environmental conditions] and a very dense urban environment.”
SUSS’ Theseira believes, however, that the question of safety goes further than the rate of accidents. “It is not practical to aim for fewer accidents, because in the initial years autonomous vehicles will be deployed with non-autonomous vehicles.” The potential for accidents rises as the autonomous vehicles share the space with other vehicles and pedestrians, with less predictable movements, he says.
“If you can’t have zero accidents, then what kind of standards are you willing to accept? Different countries will have different jurisdictions and testing regimes before they would accept autonomous vehicles,” Theseira adds.
Driverless or drive less?
The KPMG study also found that, importantly, people in Singapore are very accepting of autonomous vehicles. To be sure, responses in the study might differ from true attitudes. Theseira reckons people are likely to be more comfortable being in some degree of control of a moving vehicle. “At the moment, I don’t think we are likely to see a move from fully human-controlled vehicles to fully computer-controlled vehicles, without intervention, in one step. What seems to be far more likely and already happening is the gradual encroachment of more and more advanced driver systems into everyday production vehicles,” he says.
This gradual process is important, Theseira says, in familiarising the public with the autonomous systems. “Today, a lot of production vehicles have automated cruise control and lane-keeping systems, which, if the driver lets them, are capable of doing a lot of work on the expressway. That gives the ordinary driver an example of what the technology can do. It seems more likely that the technology will grow in scope and give drivers less and less to do, until one day, there is no conceptual difference for the driver between being driven by the computer and having a lot of assistance,” he says.
KPMG’s Ramamurthy agrees, and gives the analogy of flight, where a significant amount is done by computers. Pilots take over during takeoffs, landings and uncharted occurrences. “We are not going to see the complete removal of human intervention very soon. [That’s needed] fundamentally to deal with some of those exceptional [situations],” he says.
Just as we are not quite expecting planes to fly on their own, one major issue that planners of an autonomous vehicle future should be taking into account is the fact that an entire subsection of jobs will become obsolete if computers take over completely. There is, of course, the assumption that drivers will be reskilled and take on different roles.
“Ultimately, it is about making choices, from a societal point of view. That has to be made in each environment,” Ramamurthy says. “If you have a workforce in this area that is largely imported or foreign, then the impetus to pick up autonomous vehicles is stronger. But if the workforce mix is more local, and… reskilling that workforce is going to require time and investment, then society is likely to move a bit slower with this issue, to deal with the consequences of that.
“That’s where I think it is not about technology but what kind of future you are trying to build for your people.”
Volvo’s Forsbergh reckons there will always be a role for humans, albeit different ones as technology advances. It could be more of managing logistics, for instance, rather than steering a vehicle. “We are approaching this in a very [broad] way and will see where the market will take this,” he says. “But still, it is important to remember that [for] quite many years to come, there will still be a driver in the truck.”
This story first appeared in The Edge Singapore (Issue 840, week of July 23). Click here to subscribe