Perhaps no one expected that self-driving technology would appear so quickly in the daily lives of ordinary people.
In 2021, relevant policies continue to open up, Robotaxi trial operation services appear in many parts of the country, and some car companies closely follow industry trends and launch mass production models with certain levels of automatic driving functions. The autonomous driving industry rejoices as technology continues to approach applications.
However, autonomous driving technology is still a new thing, and due to the lack of in-depth understanding of it, people have different views:
One of them believes that the safety of automatic driving is much higher than that of human manual driving, and you can trust automatic driving technology with confidence, but because of this, many tragedies caused by human uncontrolled vehicle driving occur; the other is because of the car accident of a certain level of autonomous vehicles, and believes that the safety of automatic driving technology is far lower than that of human driving, and even has a fear of it.
In fact, most of the above views are based on only a small number of cases, which is inevitably biased.
Recently, the "White Paper on Traffic Safety of Autonomous Vehicles" jointly completed by China Automobile Center, Tongji University and Baidu was officially released. The white paper analyzes the traffic safety of autonomous vehicles by combining policies and regulations related to the traffic safety of autonomous vehicles, safety technologies, and the safety of autonomous driving and human-driven cars in the current situation of road traffic accidents in China. A large number of facts and figures support it to enhance the scientific and persuasiveness of the white paper.
New Intelligent Driving tries to unveil the true veil of autonomous driving through the data and cases in the "White Paper on Traffic Safety of Autonomous Vehicles".
Reduce eight adults for accidents
The innovation of transportation means has promoted the accelerated development of modern human society to a certain extent.
Since the birth of the world's first car on January 29, 1886, the footprint of the individual has been continuously extended with the help of the car, the circulation between the means of production has become smoother, and the automobile has also made the connection between people close and frequent.
However, in the driving process of the car in which humans have control, various factors lead to a series of car accidents at any time.
According to the World Health Organization's 2018 State of the World Health Report on Road Safety, about 1.35 million people die in road traffic accidents every year, equivalent to 1 person dying in traffic accidents every 24 seconds, and about 20 million to 50 million people suffering non-fatal injuries. Today, road traffic injuries are the eighth leading cause of death worldwide.
The current situation of traffic accidents in our country is equally grim. According to the Statistics Annual Report on Road Traffic Accidents of the Traffic Management Bureau of the Ministry of Public Security, from 2017 to 2019, china's traffic accidents occurred an average of 231,900 times a year, with an average annual death toll of 63,000, and another 240,000 people were non-fatally injured.
Through CIDAS, the largest road traffic database in China, the White Paper on Traffic Safety for Autonomous Vehicles screened out 6967 accidents caused by passenger cars from 2011 to 2021, and found that about 81.5% of passenger car accidents were caused by human factors of drivers. Among them, about 79.9% are due to the subjective error of the driver caused by the accident, and about 20.1% is due to the limited ability of the driver to cause the accident.
Specifically, the detailed causes of accidents caused by subjective errors by drivers can be divided into failure to give way according to regulations, excessive speed, illegal use of lanes, drunk driving, violation of traffic signals, and fatigue driving. Among them, the proportion of failure to give way according to regulations is as high as 43.4%. The causes of accidents due to limited driver capabilities can be subdivided into failure to pay attention to the behavior of other traffic participants, failure to maintain a safe distance, low driving proficiency, and other wrong driving behaviors.
Observing the above causes of passenger car accidents, it can be found that most of the accident causes are closely related to the violations of "people". If autonomous driving technology replaces human drivers, the traffic accident rate may be significantly reduced.
According to a study by the foreign agency Eno Centre for Transportation, if 90 percent of cars on U.S. roads were converted to self-driving cars, the number of crashes would fall from 6 million to 1.3 million and the death toll from 33,000 to 11,300.
Part of the reason for this data is based on the real-time dynamic acquisition and recognition of the surrounding environment by self-driving cars through vehicle-road collaboration technology and multi-sensor fusion, and then take corresponding measures against other traffic participants in a timely manner.
Based on CIDAS database data, the following figure shows the distribution of pedestrian centers of gravity with a detection range of 30°, 45° and 60° in the 1.0 seconds before the collision of vehicles controlled by human drivers and autonomous driving, respectively. Compared with the limited ability of humans to drive cars, the automatic driving perception function can detect more than 90% of the situations that may lead to accidents in advance.
In addition, the self-driving car that is not affected by fatigue and emotions, its relatively perfect decision-making planning can ensure that the vehicle is driven in accordance with the traffic regulations, which means that the normal driving of the autonomous vehicle can completely reduce the subjective error of human driving, that is, 80% of the accident causes of human driving can be effectively avoided, and traffic safety will be greatly improved.
However, self-driving cars cannot completely eliminate accidents and achieve 100% driving safety.
In October 2020, Waymo released the Waymo Public Road Safety Performance Data white paper, which for the first time conducted a public discussion of collisions encountered by self-driving cars. The report shows that Waymo's self-driving cars have had a total of 18 collisions or minor contact accidents during the approximately 9.81 million kilometers of real-world road driving.
In addition, waymo's self-driving cars had 29 collisions in simulated environments, mainly due to non-compliance by other road traffic participants.
Based on the above data, it can be seen that the safety of automatic driving is much higher than that of humans, but it does not mean zero accidents.
Driver safety is one more barrier
Since autonomous driving cannot achieve 100% safety, and human drivers still need to pay attention to the road surface and intervene in time, what is the significance of introducing autonomous driving into the car?
The answer is that autonomous vehicles improve traffic safety when driving on the road.
When the penetration rate of self-driving cars reaches 25%, 50%, 75%, and 100%, traffic conflicts will be reduced by 12%-47%, 50%-80%, 82%-92%, and 90%-94%, respectively.
Although traffic accidents are difficult to avoid, self-driving cars can perceive potential risks in advance through vehicle-to-road collaboration technology and multi-sensor fusion, and take corresponding countermeasures in time, thereby alleviating accident casualties and reducing the risk of collisions in secondary accidents.
At present, in the process of several years of self-driving car testing by many companies around the world, autonomous vehicle-related accidents are only occasional cases. By the end of 2020, among the enterprises with a road test mileage of more than 2 million kilometers in Beijing, Baidu Apollo's test vehicles and test mileage were the first, and there was no adverse impact on other traffic participants during the test.
In addition to reducing traffic accidents caused by the outside world, autonomous vehicles are designed to be safety-oriented and ensure the personal safety of passengers through safety systems.
The safety system of autonomous vehicles can be roughly divided into three layers of safety system: main safety system, redundant system, and remote cloud driving.
In terms of the main safety system, the self-driving car is equipped with a highly stable, real-time, low-latency operating system, and its reaction time is much lower than the 250 milliseconds of the human driver, and the self-driving car can respond in time to respond to dangerous situations and reduce the possibility of accidents.
In addition, the environmental perception ability provided by its multi-sensor, combined with high-precision maps and high-precision positioning of road and vehicle information, can cross-verify vehicle prediction information and road actual information to ensure the reliability and safety of autonomous vehicles.
Autonomous vehicle perception under the influence of extreme weather
In terms of redundancy systems, the computing units of autonomous vehicles and hardware such as domain control, sensors, and vehicle chassis are fully redundant. When the abnormal state of software and hardware is detected or the ODD response capacity is exceeded, the autonomous vehicle will launch the redundant system in a targeted manner, and the automatic driving function will be immediately downgraded, and the driver can completely disable the automatic driving system to safely control the vehicle.
Remote cloud driving, on the other hand, provides operators with the ability to remotely control autonomous vehicles in real time. When the self-driving car is difficult to automatically disengage or complex operation assistance, a single operator can remotely control the vehicle in the case of 5G network, so that it can quickly move away from trapped or dangerous areas to ensure the personal safety of passengers.
Through more accurate environmental perception and higher safety protection capabilities, self-driving cars can improve traffic safety and reduce casualties.
Industry standards range from disorder to order
With the continuous improvement of the degree of intelligence of automobiles, the driving assistance function and even the automatic driving function have almost become the standard configuration of future mass production models. Higher technical requirements threaten the survival of OEMs with insufficient investment in research and development, and assisted driving functions to be improved. In this case, in order to flaunt the superiority of their respective products and stabilize market share, some OEMs have declared their L2-level assisted driving functions in marketing, which are higher-level L4 and even L5-level automatic driving functions.
Under the overwhelming publicity of some main engine factories, assisted drivers are forced to put on the cloak of "automatic driving", and some users highly trust the so-called "automatic driving" function under the intensive propaganda offensive, which has led to a human tragedy. At this time, some OEMs quickly tore off their disguises and unjustly claimed innocence. It is precisely for this reason that the autonomous driving industry, which is relatively cautious about promoting technology, is collectively infected with haze.
The release of the "White Paper on Traffic Safety of Autonomous Vehicles" plays a role in clearing the source, and the automatic driving industry urgently needs such objective and scientific scientific information to pass on the correct knowledge of automatic driving.
When users know how to distinguish between assisted driving and autonomous driving, and when to correctly enable assisted driving, the related accidents will be reduced accordingly. In a better public opinion environment, the technical application of autonomous driving may be smoother.
On the other hand, the White Paper on Traffic Safety of Autonomous Vehicles elaborates on the strict requirements of the policy for the safety of autonomous vehicles, as well as the multiple safety redundancies of autonomous vehicles for autonomous vehicles by self-driving players including Baidu Apollo.
Autonomous vehicles with safety as the first nature, at the beginning of vehicle design, follow the requirements and design ideas of functional safety and expected functional safety, and improve the safety of autonomous vehicles through core algorithm strategy design, hardware and software redundant safety design, remote cloud driving technology, and full-process test and verification technology.
From system design to actual operation, self-driving cars have built a three-layer safety system, which is far more complete than human driving. In this way, the release of the "White Paper on Traffic Safety of Autonomous Vehicles" can dispel people's worries and doubts about the safety of autonomous vehicles to a certain extent, prompt more ordinary people to experience autonomous driving services, and thus reduce the cost of promoting autonomous vehicles.
More importantly, the White Paper on Traffic Safety of Autonomous Vehicles does not deliberately avoid accidents related to autonomous vehicles, but truthfully lists and carefully analyzes the causes of accidents. Through real events, the objective facts that prove that automatic driving is not completely safe reduce the psychological expectations of ordinary people for automatic driving technology.
Uber self-driving car vs. two-wheeler accident
To some extent, the active disclosure of autonomous driving accident cases has also made the public more tolerant of automatic driving, rather than causing the public to blame technical defects or cause unnecessary technical panic after related events.
summary
In the "White Paper" released this time, it was also disclosed that a Baidu Apollo vehicle encountered a traffic accident because of the illegal driving of a human driver and ran a red light.
It is reported that in the incident, Baidu Apollo's test vehicle drove normally in accordance with traffic laws, and a social vehicle suddenly ran a red light to face the test vehicle, and eventually led to a collision between the two vehicles.
If we delve into the cause of this collision between an autonomous vehicle and a human-driven vehicle, the reason is that the illegal driving of human drivers undermines traffic safety, resulting in collisions. Excluding this variable, autonomous vehicles will not have such accidents caused by violations.
After the superposition of multiple autonomous driving-related accidents, the industry has experienced a round of reflection on the definition of "automatic driving", and the society has further understood what automatic driving is.
However, the "wave of reflection" that main engine manufacturers have erupted on the Internet is both rapid and short-lived, and its influence is beyond reach and unsustainable, and it is impossible to systematically change the public's misconception about automatic driving.
An autonomous driving white paper jointly completed by a number of scientific research institutions and industry leaders, explaining automatic driving by combining policies, technologies, applications and other aspects, is the best carrier for the true face of autonomous driving that has also been "demonized", and its influence and authority are not the same.
The "White Paper on Traffic Safety of Autonomous Vehicles", which is jointly worked together by China Automobile Center, Tongji University and Baidu, has been released, and its positive effect on the autonomous driving industry may become more and more obvious at the time when autonomous driving technology tends to a new stage of technology application.
Autonomous driving technology is still steadily moving forward, and it is expected that more things like the "White Paper on Traffic Safety of Autonomous Vehicles" will appear to clean up the dirt for the industry, restore the true appearance, and promote the benign development of technology.
Leifeng networkLeifeng networkLeifeng network