Today, Baidu Apollo's autonomous driving test mileage has exceeded 25 million kilometers, unmanned tests have also driven from "Shougang Park" into the social roads in the core area of Yizhuang, and commercial toll operations have been applied from the park to the public's daily travel services.
China's self-driving "Whampoa Military Academy" is heading for a higher level of unmanned autonomous driving, and the biggest stumbling block on the road is China's complex traffic conditions itself.
Domestic urban roads are very complex, bringing higher challenges to driverless technology. Taking Beijing as an example, the density of urban road participants in Beijing is 15 times that of California, the United States, allowing autonomous vehicles to enter the daily life, it has to face countless unexpected scenarios, how to deal with these scenarios to achieve safety higher than human drivers, Baidu Apollo gave a complete set of logic.
The premise of no one is safety
At the Baidu Apollo Open Day, Xu Baoqiang, director of Baidu's automotive robot department, elaborated on the safety methodology of automatic driving.
With "safe" and "known" as the horizontal and vertical coordinates, divided into four quadrants, it is committed to continuously improving the ability of bicycle intelligence to increase safe, so that unsafe can be pushed to safe; on the other hand, the use of monitoring redundancy to increase the detection and response mechanism of electronic and electrical system failures will push unknown to known. The final is the first quadrant of "known-safe" that continues to expand, compressing the "unknown-unsafe" scene.
In order to achieve the final safety effect of unmanned driving, Baidu Apollo proposed a three-layer safety system design.
In the main system security, Baidu Apollo's fully self-developed real-time on-board OS can make the vehicle reflect faster than the human driver; the pan-perception system equipped with it is redundant to ensure accurate detection ability and build display change recognition and response capabilities; the decision-making and planning system to ensure that the vehicle makes behaviors that conform to the right of way, and at the same time complete theoretical calculations to ensure the safety of vehicle behavior; in addition to bicycle intelligence, vehicle-road collaboration can independently provide over-the-horizon perception of complete closed-loop capabilities, such as non-motorized vehicles, pedestrian ghost probes, construction roads, Monitoring such as low obstacles is also a supplement to the perception of bicycle intelligence.
Another layer of safety monitoring and redundancy system, to achieve full redundancy of hardware and vehicles, the failure of any single component of the suite does not affect safety, and based on the improvement of monitoring and problem detection, design a complete downgrade and functional safety strategy.
Finally, Baidu Apollo also adopts the 5G cloud driving method to carry out "one-to-many" control of autonomous vehicles running on normal roads, which gives full play to the advantages of domestic infrastructure and accelerates the landing and popularization of unmanned driving.
In addition to the functional safety methodology and the three-layer security system, Xu Baoqiang also introduced the necessity of test verification. With simulation-driven rapid verification, more than 90% of the new collision problems can be repaired every six months, so it is also essential to complete the test and verification system for unmanned people.
At the event site, Xu Baoqiang called the team that was undergoing daily unmanned driving tests through the way of live connection, and it can be clearly seen from the live broadcast screen of jiugongge that unmanned vehicles are driving on Yizhuang road, the traffic environment is the most complex, and the intersection with the highest density of people and traffic can also be easily coped with.
Based on the three capabilities of safety methodology, three-layer safety system, and test and verification system, Baidu Apollo believes that the comprehensive safety of automatic driving has been higher than that of human drivers.
Radish is about to run out of the results
The most real feeling to the people is the daily operation of Radish Run Robotaxi, in Beijing's Yizhuang Economic Development Zone, There are more than 600 sites of Radish Run, covering most of the areas where residents live.
Wei Dong, vice president and chief safety operation officer of Baidu Intelligent Driving Business Group, introduced that after the release of the Radish Run brand, in the half year of Yizhuang manned test demonstration operation, the number of vehicles, site density, and order density increased by more than 50%, and the average daily peak of bicycles reached 28.
At the same time as the large-scale unmanned fleet test is opened, radish fast running is also accelerating the commercial operation landing.
In the fourth quarter of 2021, the number of orders for radish fast running manned reached 213,000, nearly doubling from the previous quarter, and more than 300,000 orders were obtained in the third and fourth quarters, and this data is still growing at a high speed in 2022. Up to now, Radish Run has begun to commercialize paid operation services in Beijing, Chongqing and Yangquan.
Wei Dong said that after the turnip run began commercialization, the user scale, frequency and stickiness showed steady growth, and the number of people who repurchased more than 4 times accounted for 30%, and the number of short-distance connections was more.
Although the radish fast run has implemented commercial charges, there are still large subsidies, and the final ride price is still lower than that of online ride-hailing cars and taxis, so after charging, the user's willingness to ride has not declined. During the street visit, some passengers said, "Because of the cheap ride on this car, if the price increases, I don't know if I will sit", and some passengers said that "price is not the first element to consider, if I can meet my car needs, I will sit."
In terms of service, most people also reflect the problem of difficulty in calling a car during peak hours, to which Wei Dong responded that this is related to the substantial growth of ride users, and Baidu Apollo will also strive to expand the scale of robotaxi and improve the vehicle scheduling system.
On the background management end of Radish Run, Baidu Apollo has built a management system around large-scale operation, including the order visualization background and the fleet inspection background, to analyze and feedback the vehicle operating conditions and service quality.
Baidu Apollo also announced that it took the lead in launching the digital yuan payment function in Beijing Yizhuang, realizing the first combination of autonomous driving and digital currency. In addition to vehicle driving technology, Baidu Apollo is also exploring technological changes that lead lifestyles and convenient travel services.
Che Yun summary
The theme of last year's Apollo Day was "Rushing to No Man's Land", and this year's unmanned team has been on the road; this year's theme "The Road to the Future" is the vision and determination of Baidu Apollo to accelerate the popularization of autonomous driving.
Nearly a decade of exploration of autonomous driving, self-driving cars from concept to reality.
Baidu Apollo said that just like 2021, it is difficult for everyone, but there are still many people who are silently studying and pursuing excellence. Today, Baidu Apollo has a new challenge in the new scene, but it is still like a "good student" who continues to learn, with the belief of pursuing universality and driving to the future.