4 hours found Baidu Apollo, NVIDIA DriveAV 561 faults
Editor’s note: This article is from WeChat public account “ New Wisdom” (ID: AI_era).
Source: sciencedaily, arXiv
[New 智元导读] Musk once said: “Automated driving will never be perfect, but 10 times safer than human driving!” Is that true? Researchers at the University of Illinois at Urbana-Champaign developed a failure assessment technique that found 561 critical safety faults in four hours in a test of Baidu Apollo 3.0 and NVIDIA’s proprietary autopilot DriveAV.
Tesla CEO Musk once said: “Automated driving will never be perfect, but 10 times safer than human driving!”
However, is that true?
A research team at the University of Illinois at Urbana-Champaign analyzed the safety reports from autopilot companies from 2014 to 2017 (covering 144 self-driving cars with a total of 1,116,605 miles) and came to the opposite conclusion:
In the case of the same mileage, humans are more than 4000 times less likely to drive an accident than autonomous vehicles.
And recently, this research team developed a fault assessment technology for autonomous driving. In the test of Baidu Apollo 3.0 and NVIDIA’s proprietary autopilot DriveAV, 561 were discovered in just 4 hours. Critical safety failure!
This research team is committed to using artificial intelligence and machine learning to improve the safety of autonomous driving technology through software and hardware improvements.
“Using AI to improve autonomous vehicles is difficult due to the complexity of the electrical and mechanical components of the vehicle, as well as changes in external conditions such as weather, road conditions, terrain, traffic patterns and lighting.” Ravishankar K., Professor of CSL Labs, University of Illinois. Iyer said, “We are making progress now, but security is still a major issue.”
