Self-driving cars have not had a great deal of favourable press since a Tesla Model S crashed in China 2016, killing its occupant, and US authorities are investigating more than a dozen cases involving arguably the world’s best-known electric automobile.

Nevertheless, Singapore’s Centre of Excellence for Testing & Research of Autonomous Vehicles (Cetran) is undeterred by the early mishaps, saying that self-driving vehicles are a key part of a plan to make the city-state “car-lite”, potentially reducing traffic accidents and keeping the roads safer for all.

As an element of that vision, fleets of autonomous vehicles may one day function as on-demand shuttle services between people’s homes and workplaces, and as public transport nodes to help improve so-called “last-mile” connectivity – the part of a journey between the nearest railway station or bus stop and its final destination.

Niels de Boer, Programme Director at Cetran says that not only can autonomous vehicles contribute to road safety and reduced reliance on private cars, but “given the manpower constraints in the transportation sector, we can harness [them] to alleviate labour shortages.”

Testing times

To support the development of autonomous vehicles, Cetran, which is run by Nanyang Technological University, conducts physical tests as part of the approval process for on-road trials. The test centre, built to simulate Singapore’s public roads, includes features typical of the urban environment, such as zebra crossings, speed humps and car parks.

Cetran also uses a digital twin of the test track to create virtual obstacle courses.

“We can stimulate digital road hazards such as vehicles emerging from junctions and children running onto the roads without having to coordinate real-life logistics,” de Boer says. That not only helps to cut costs but also expands the types of testing Cetran can perform without any real-world risk. The centre has tested more than 50 autonomous vehicle projects so far.

Two years ago, Nanyang Technological University and Volvo Buses launched the world’s first full-sized, autonomous electric bus. At 12 metres long and with a full capacity of close to 80 passengers, this project was a standout for de Boer. “It’s impressive that a vehicle this big can run autonomously on the tight corners in our test track,” he says.

Unlike ordinary buses, the autonomous electric bus produces no carbon emissions from its engine and uses 80 per cent less energy than a diesel bus of equivalent size.

The bus is also equipped with GPS, which runs in parallel with an inertial management unit that measures speed and movement. This improves the bus’s navigation when it traverses uneven terrain and makes sharp turns, ensuring a smoother ride for passengers.

Another Cetran autonomous vehicle project involves road sweepers. Unlike most vehicles, road sweepers are designed to follow kerbs and move dynamically from one area to another, so the organisation had to reframe its testing approach.

Cetran has collaborated with the automation wing of German conglomerate Siemens to develop a vehicle-to-infrastructure communication system that allows road hardware such as traffic cameras, traffic lights and parking meters to share information with vehicles.

“For example, we can place sensors at the top of a lamppost to monitor road conditions and wirelessly send this information to autonomous vehicles nearby,” de Boer says. This allows self-driving vehicles to make better decisions as they receive data on potential hazards they would not otherwise have been able to capture.

Dynamic driving

The bulk of Cetran’s work lies in developing autonomous vehicle requirements, which is a challenge because driving in Singapore is highly regulated.

“The process of getting a driving licence is tedious, regular inspections on vehicles are mandatory, and the Land Transport Authority watches over the car modification scene like a hawk,” de Boer says.

Autonomous vehicles present a number of other challenges. For example, mirrors are vital on any normal vehicle, but they’re unnecessary on an autonomous one. “Instead, what we need are better cameras as driverless cars have computer vision and their field of sight is more comprehensive,” de Boer says.

And as traffic conditions are ever-changing, trying to govern every aspect of road users’ interactions is practically impossible. For instance, Singapore drivers learn that they have to turn on their headlights after 7pm for better road visibility. But what should autonomous vehicles do if it is already dark before 7pm, or still bright out after 7pm? Sticking strictly to a formula is not ideal because conditions change.

Then comes the duty of care expected of all road users. Motorists are obliged to ensure that any action they take does not cause injury to others or damage to property. “A driver can complain about an illegal pedestrian crossing, but that still doesn’t give them the right to hit a person,” de Boer says.

Autonomous vehicles need to be able to make distinctions and not just stick to fixed rules. To navigate such a dynamic environments, Cetran has taken a watch-and-learn approach. As it conducts safety checks, it learns what is most important to regulate and control.

One example of this involves the quality of sensors, which have to detect pedestrians and other vehicles from certain distances. “They also have to be adequately responsive to lane markings, traffic signs and traffic lights,” de Boer says.

Tesla may be in hot water over the safety of its self-driving technology, but Cetran is working steadily to make sure that the systems behind the next revolution in transport are sufficiently developed to avoid anything like a string of similar incidents in Singapore.