Data from more than 2,000 self-driving vehicles has contributed to a study concluding they may be safer than humans in some conditions, and potentially more dangerous in others.
A paper published in Nature Communications this week found that autonomous vehicles are less likely to be involved in accidents than humans in everyday scenarios including keeping in the same lane and maintaining speed with the flow of traffic.
The study, which collected data from 2,100 autonomous vehicle and 35,133 human-driven vehicle accidents between 2016 and 2022, also found self-driving cars are likely to avoid rear-end and sideswipe accidents, which were 0.5 and 0.2 times less common respectively.
Led by Shengxuan Ding, a PhD student at the University of Central Florida, the study found that autonomous vehicles could possibly improve road safety, but there are also areas that need improvement.
For example, the research found self-driving vehicles were on average 5.25 times more likely to have accidents in low-light conditions – during the early morning or as the light was fading at dusk. They were also 1.98 times more likely to make a mistake than humans while executing turns. The researchers said their work used more data from both autonomous and human-driven vehicles than earlier studies.
- Waymo issues software fix after driverless taxi hits telephone pole
- Tesla's Autopilot false advertising tussle with California DMV must go to trial
- Chinese car brands hit accelerator on road tests for level three autonomous driving tech
- Researchers warn robot cars can be crashed with tinfoil and paint daubed on cardboard
It said the poor performance of autonomous vehicles in low light conditions might be attributed to "a lack of situational awareness in complex driving scenarios and limited driving experience."
Improving autonomous vehicle performance in such conditions might involve "advanced sensors, robust algorithms, and smart design considerations."
"Key strategies include enhancing weather and lighting sensors, implementing redundancy measures, and integrating sensor data effectively. By focusing on these aspects, the safety of [autonomous driving systems] can be significantly enhanced in challenging scenarios," the research said.
One thing to note here is that the study covers data up to 2022. In the past two years, autonomous driver-less taxi cabs operated by Google-owned Waymo, at least, have ferried around more and more people, and can handle trips at night and during the day. They're not perfect, and are limited to a few cities, but it's safe to say there have been at least some improvement in the past 24 months or so.
Indeed, investors have made big bets on them.
Ark Investment Management said Tesla stock was its largest holding and suggested the value of the Musk-led biz could increase by $2,000 per share – or 1,022 percent by 2027 on the back of its potential developments in autonomous driving.
CEO Cathie Wood recently said Tesla stock was the greatest opportunity in AI and the automaker would build the most advanced autonomous driving software available. ®