A BUMPY ROAD AHEAD
In October last year, a self-driving car from nuTonomy collided with a human-driven lorry while undergoing tests in the one-north district, in what was believed to be the first mishap in Singapore involving an autonomous vehicle.
While no one was hurt in the incident, it sparked concerns about the safety of driverless technology.
Following investigations, nuTonomy said the accident was caused by “an extremely rare combination of software anomalies”, which affected how the vehicle detected and responded to other nearby vehicles when changing lanes.
It has since improved its software system and resumed trials without incident. Earlier this month, nuTonomy announced a partnership with automaker Groupe PSA in France to install its sensors and computing systems in customised Peugeot vehicles. This will provide seamless integration of AV software with current vehicle hardware.
There have been a few reported accidents elsewhere.
In May last year, 40-year-old Joshua Brown was killed in Florida after his Tesla, which was on “Autopilot” – a driver-assist mode with automatic acceleration, steering and braking functions – smashed into a tractor trailer.
Safety regulators later cleared Tesla’s “Autopilot” system of fault, and in their findings said that Mr Brown ignored Tesla’s warning to maintain control even while the car was in such a mode.
Tesla said the driver and the car’s Autopilot failed to distinguish the white side of the trailer against a bright sky in the accident.
The firm said it has improved its systems, with greater reliance on radar than on cameras, and has incorporated safety features to disable Autopilot when drivers are not paying attention.
THE FUTURE IS HOPEFUL
Despite these incidents, developers and researchers of self-driving cars believe that when the technology is fully mature, these vehicles will be safer than human-driven ones.
This is because of the untapped potential of artificial intelligence, particularly in the area of “deep learning”, in which computers mimic a human brain’s ability to recognise patterns in huge data sets and make predictions.
This will allow self-driving cars to better deal with the myriad scenarios on the roads, such as deciding when to drive through a road intersection without traffic lights, for example, or predicting the behaviour of other motorists and pedestrians.
Researchers also expect the hardware which helps autonomous vehicles to “see” to become more sophisticated.
Currently, self-driving technology relies on Lidar (Light Detection and Ranging) sensors and smart cameras. Lidar uses laser beams to map out the surrounding environment and objects, while the camera scans for traffic lights, road markings and pedestrians.
The technology is still evolving – two months ago, scientists at the Nanyang Technological University (NTU) unveiled an ultra-fast camera that mimics the human eye.
It can track minute changes in the scene viewed, even in low light, and has applications for self-driving cars and drones.
As technology matures, the LTA is also ramping up its regulatory oversight.
A 1.8ha AV test circuit, located just outside NTU, is expected to be ready in the second half of this year. The facility will be used to test and certify self-driving vehicles in a simulated environment.
The test circuit, which is part of a Centre of Excellence for Testing & Research of Autonomous Vehicles (Cetran) funded by the Land Transport Authority (LTA) and JTC, in partnership with NTU, will have slopes, roundabouts and a section which simulates rainfall.
THE CASE FOR AVs
Globally, around 1.3 million people die in traffic accidents every year. According to studies, over 90 percent of road accidents are the result of human error arising from poor decision-making, distraction and fatigue.
Scientists tout driverless technology as the antidote to this, as computers never get weary or distracted and, more importantly, are not prone to road rage or erratic behaviour.
These are solid grounds for why money and effort should continue to be poured into the research on self-driving technology, despite the occasional setbacks.
The question then is not whether AVs are the future, but when they can fully replace human-driven private cars, buses, taxis and commercial vehicles. Optimists reckon that this can happen in a decade, but more conservative camps say that it will take at least another 50 years.
But when the time comes, will people be ready for the change?
A Deloitte study of 17 countries – including the US, Germany, China, and Japan – found that consumers are still sceptical of the safety of AVs. Published in January, the report found that as many as 79 percent of consumers in Japan believe that fully autonomous vehicles will not be safe.
Researchers and automakers thus have an uphill task in getting the buy-in of the public to put their lives in the hands of self-driving vehicles.
Related story: Bumps along a driverless road
Related story: Time to rev up discussion about insurance for driverless cars