Autonomous Vehicle Traffic Safety White Paper: Autonomous Driving Reduces Traffic Accident Rates

Self-driving car traffic safety white paper courtesy of Baidu

On December 16, the "White Paper on Traffic Safety for Autonomous Vehicles" (hereinafter referred to as the "White Paper"), compiled by China Automotive Technology and Research Center Co., Ltd. and Tongji University and Baidu Apollo, was officially released, putting forward the view that "automatic driving is safer than human driving, but it is not a 'zero' accident".

The white paper points out that people are the most uncertain factor in traffic safety. According to the CIDAS (China Traffic Accident In-Depth Survey) database, the driver's human factor accounted for about 81.5% of the 5664 accidents in which passenger cars were involved from 2011 to 2021. Among them, because the driver cannot identify and perceive the danger in advance, the subjective error of the driver causes 79.9% of the accidents. Accidents due to failure to give way according to regulations accounted for 43.4%, followed by excessive speed, illegal use of lanes, drunk driving, violation of traffic lights and fatigue driving.

Compared with human driving, the white paper proposes that autonomous driving is safer from three aspects.

First of all, automatic driving relies on a full range of perception systems, intelligent decision-making systems and accurate execution systems, which do not require driver operation, and can eliminate dangerous behaviors such as fatigue driving and drunk driving in human driving. At the same time, strictly abiding by traffic rules can prevent violations of traffic rules in human driving.

Secondly, autonomous driving can pre-perceive the surrounding environment from a full perspective through the fusion of multiple sensors and vehicle-road collaboration technology, which can avoid risks in advance and effectively reduce the incidence of accidents.

Finally, self-driving cars have the ability to learn and continuously upgrade, based on rich travel scenarios can truly solve the problems of rigid need, pain points and high frequencies in travel, improve the trust mechanism of all participants from the source, and make all nodes of travel "traces".

However, the white paper also emphasizes that autonomous driving and zero accidents cannot be equated. Autonomous vehicles are in a dynamic and variable traffic environment, and in the face of complex and sudden danger scenarios, the failure of autonomous vehicles themselves to deal with effectively, other traffic participants and other factors are also important causes of autonomous vehicle accidents. (End)