Safe driving of autonomous vehicles tests social governance capabilities
◆Open road tests in many places and start demonstration applications ◆Partially solve the problem of fragmentation and decentralization of supervision ◆Need to establish new types of autonomous driving road traffic insurance ◆Revise legal uniform standards to promote smart transportation
Our reporter Zhang Hao
"Automatic driving begins." After pressing the autopilot button on the steering wheel, the driver of the Xiaoma Zhixing autopilot test car placed his hands on his knees and looked ahead.
Such vehicles are not uncommon on open roads in Beijing's autonomous driving test area.
In recent years, many cities in China have set a "runway" for autonomous vehicle testing, and a small number of cities have tried to open Robotaxi (translated as autonomous taxi) manned trial operation test.
In order to create a good testing and demonstration environment and promote the healthy development of the intelligent connected automobile industry, the Ministry of Public Security and other ministries and commissions have recently jointly revised the "Management Standards for Road Testing and Demonstration Application of Intelligent Connected Vehicles (Trial)" (hereinafter referred to as the management standard) and published it solicit opinions.
Can road traffic rules involving autonomous driving meet practical needs?
How to improve the level of relevant social governance capabilities?
A reporter from the Rule of Law Daily interviewed scientific and technological personnel of R&D enterprises and experts on related legal issues.
"Knowing the truth"
The reporter rides in a test car that has obtained a T3 level autopilot test license.
After the test driver left the steering wheel, the steering wheel began to rotate automatically.
Proceed to Yuncheng Street in Daxing District. According to the route planning, vehicles need to turn left.
The intersection ahead is a "non-protected left turn" intersection, that is to say, left turn and straight traffic lights are not distinguished or differentiated but let go together.
When going straight or turning left, avoid vehicles coming from the opposite lane and pedestrians crossing the road. This is a test for autonomous vehicles.
There is still about 50 meters away from the intersection and the green light is 5 seconds away. A tricycle on the sidewalk is passing through a red light at low speed, and an oncoming straight vehicle may have to run a yellow light.
The reporter saw that the self-driving car seemed "knowledgeable": when it decelerated steadily to the green light and passed the stop line, the tricycle had completely passed the intersection; the vehicle that gave way to the yellow light immediately started and passed the intersection.
Self-driving cars are more accurate than natural drivers when turning.
When the vehicle is driving to another unprotected left-turn intersection, neither the front nor the left vehicle can accurately enter the straight lane after turning left, and the autonomous vehicle is aligned with the lane to be entered.
On the two display screens of the center console of the self-driving car, one displays the map and route, and the other displays the objects and pedestrians around the vehicle.
Xiaoma Zhixing’s engineers told reporters that this self-driving test vehicle combines a full-stack L4 level autopilot software and hardware system with a roof sensor device and in-vehicle components (L1-L5 are driving automation levels, and the highest is L5 fully automated, L4 (Highly automated), it can “see clearly” objects within 200 meters of the surrounding area, predict pedestrian and vehicle behavior without human intervention, and make real-time driving decisions such as avoiding, overtaking, and changing lanes, by checking the steering wheel, brakes, throttle, etc. The precise control to complete the driving behavior.
Also "road test"
"Like natural human drivers, self-driving test vehicles must also participate in the'road test' at the simulated venue." Xiaoma Zhixing's engineer said that after passing multiple subject tests corresponding to its intelligence level, the self-driving car can obtain a license plate. Test on the road.
After inquiries, the "Beijing Autonomous Vehicle Road Test Capability Evaluation Contents and Methods (Trial)" classified the autonomous driving test vehicles into 5 levels, with T5 being the highest.
The T3 license plate automatic driving test vehicle needs to learn traffic signs and signals other than command gestures, and have skills such as changing lanes, passing bus stops, pulling over, and turning around at intersections.
There were not many vehicles on the road when the reporter tried.
The 30-minute autopilot process was very smooth, with no emergency stop or sudden start. The two mergings to the left side slowed down and waited for the straight-going vehicle to pass through instead of accelerating and congesting.
This little horse Zhixing test vehicle has followed the T3 level specifications and traffic regulations guidelines and completed the automatic driving action.
According to Beijing’s relevant regulations, compulsory tests for natural persons such as overtaking, reversing and entering the garage, side parking, over-limit wide doors, narrow road U-turns, etc. are compulsory courses for T4 license vehicles. Actions such as slippery roads and passing abandoned roads are necessary skills for T5 license plate vehicles.
It is understood that as of early January this year, the Beijing test vehicle license plate has just been issued to the T4 license plate, and the T5 license plate has not yet been issued.
Turn on the "green light"
The reporter interviewed a number of companies and experts in the field of autonomous driving and learned that driverless test vehicles have been tested on the roads in many cities, and some have been put into trial operation tests of Robotaxi and autonomous buses.
"Colleagues who live in Yizhuang, Beijing can book the company's driverless taxis with zero cost for commuting," said the engineer of Xiaoma Zhixing.
In Nansha District, Guangzhou City, Guangdong Province, Xiaoma Zhixing landed in China's first normal operation Robotaxi fleet in 2018, which is mainly open to employees and invited users.
Recently, on a semi-open road in the Shougang Ruins Park in Shijingshan District, Beijing, the reporter saw Baidu self-driving buses and passenger cars being tested.
According to the relevant person in charge of Baidu, Baidu's autonomous driving travel service Apollo GO has been tested in Beijing, Changsha, Hunan, and Cangzhou, Hebei. Among them, Changsha has opened the road driver to remove the driver's driverless test license.
At present, Baidu has obtained more than 200 national licenses, including the first five T4 licenses in Beijing.
Tencent has L2 to L4 autonomous driving capabilities and has obtained L4 autonomous driving road test licenses in Beijing and Shenzhen, Guangdong. At the same time, it has cooperated with intelligent network test bases in Changsha, Hunan, Xiangyang, Hubei and other places to build a virtual and physical test system.
Zhang Yanlai, director of the Kenting Law Firm in Zhejiang Province, combed through the testing of unmanned vehicles in 16 cities and found that at present, the types of unmanned vehicles allowed to be tested in various places include taxis, buses, and subways. The test sites cover open roads, closed roads, and education. Parks, logistics parks, closed highways, tourist resorts.
Zhang Yanlai said that in addition to the four municipalities directly under the Central Government, provincial capital cities such as Guangdong and Jiangsu have also opened up road tests of autonomous vehicles in different scenarios.
In addition, Zhenjiang in Jiangsu, Dalian in Liaoning, and Cangzhou in Hebei have also joined the road test camp.
"Guangzhou, Changsha, Beijing and other cities have begun to issue driverless test permits that do not require safety officers." Cao Jianfeng, a senior researcher at Tencent Research Institute, said that with the release of the draft management regulations, autonomous vehicles have shifted from road testing to demonstration Application stage.
Law paving
The reporter learned from Xiaoma Zhixing and Tencent that previously, autonomous vehicles had to be re-examined to obtain test licenses in various regions. The lack of interconnection, intercommunication, mutual recognition, and mutual trust mechanisms for licenses in various regions increased R&D costs.
The draft of the management norms for soliciting comments established a mutual recognition mechanism, but there is still no unified standard for the norms and opinions issued by various regions.
"The draft for comments partially solves the problem of fragmentation and decentralization of supervision, but for the further implementation of autonomous driving, especially unmanned driving without a natural person driver or safety officer, the adjustment of relevant laws and regulations is still needed." Cao Jianfeng said.
Traffic accident liability and insurance have a very close relationship with all parties. The companies interviewed all believe that new road traffic insurance products need to be established for autonomous driving.
Zhang Yanlai believes that to improve the level of social governance in the field of autonomous driving, it is necessary to further match the content related to autonomous driving in the road traffic safety law and highway law.
The relevant legal relationships between producers, operators, operators, developers, drivers, passengers and other subjects involved in autonomous driving have not been clarified, and legal responsibilities have not been clarified, and laws and regulations have not dealt with special problems in autonomous driving. Solve the liability and compensation issues in emergencies such as software and hardware being damaged by a third party.
In response, Cao Jianfeng proposed to amend the Road Traffic Safety Law and related regulations to give the autonomous driving system legal status.
At the same time, special liability and insurance regulations are formulated for autonomous vehicles, including a black box mechanism for recording driving activities.