A US government report that exonerated car maker Tesla from blame over an accident involving its Autopilot technology has been criticised in a new report from a risk-management consultancy.
The US government report was commissioned following a fatal accident in 2016. It found the driver at fault for failing to appreciate the shortcomings of Tesla's advanced driver-assistance system.
According to the new analysis by Quality Control Systems Corporation (QCSC), Tesla's Autopilot system is not error-free. In particular, it adds that the Autosteer component of Autopilot, which keeps the vehicle in the same lane, may even increase the risk of crashes.
Following the fatal 2016 crash in Florida, which involved a Tesla Model S and a tractor trailer, the National Highway Traffic Safety Administration (NHTSA) investigated the safety of Tesla's Autopilot driving system.
In its report, which was submitted in 2017, NHTSA said that the Autopilot system of Tesla's Model S did not identify the tractor trailer making a left-hand turn in front of the car from a cross street. However, the report held the car driver responsible for the crash, stating that he failed to pay proper attention to the traffic.
The report added that the Tesla's Autosteer system installed on Tesla vehicles in 2014-2016 were not just safe, but may have helped cut crash rates by up to 40 per cent.
However, the new investigation, carried out by QCSC using data obtained via the Freedom of Information Act, concludes that the NHTSA analysis was wrong. It indicated that Tesla's Autosteer system may actually be linked to more crashes.
According to QCSC, the NHTSA analysis ignored the entire mileage driven by the 43,781 cars it studied. In fact, NHTSA used mileage data for just 14,791 vehicles, which inflated the crash rate (before the introduction of Autopilot) and led to NHTSA's erroneous conclusion.
NHTSA's analysis of mileage and airbag deployment had claimed that there were 1.3 airbag deployments per million miles before the introduction of Autopilot and 0.8 deployments per million miles afterwards.
But, QCSC found that airbag deployments actually went up from 0.76 per million (before Autopilot) to 1.21 per million (afterwards), an increase of 59 per cent.
Autosteer is the lane-keeping component of the Tesla's Autopilot system. It was first introduced in 2014, but was upgraded in 2016.
Earlier this year, Elon Musk claimed that Tesla will introduce autonomous driving technology in the near future - a development that some specialists believe is more likely decades away.
Indeed, experts believe that even if Tesla were to roll out self-driving capabilities in the near future, regulations would still require drivers to be fully responsible on the road, and to be able to take control of the vehicle when needed.
The Spark #4 - 'The AppStore featured us and we had 25,000 downloads in two days, then the platform fell over'
Stuart Sumner speaks to Matthew Knippen, CEO and co-founder of Charge Running, an audio-based running app offering coaching and social aspects
Accessible technology can unlock opportunities in every part of society, the firm believes
ValueLicensing alleges that Microsoft pressurises customers not to sell unused Windows and Office licences, to the detriment of the secondhand market
Bug bounties, or paying ethical hackers to find software flaws, are common in the infosec space. Can the same be applied to AI?
Delays to the DSAB programme have cost £173m so far