How hard is autonomous driving?
Many in the industry have likened autonomous driving to a lunar landing plan. But now, Baidu Apollo is creating a "moon landing moment" in China's autonomous driving industry.
On April 28, Beijing issued a notice of unmanned manned demonstration application, Baidu became one of the first batch of approved enterprises, and its autonomous driving travel service platform Radish Run officially opened unmanned autonomous driving travel services.
This is the first time that China has allowed "no one behind the steering wheel", and within 60 square kilometers of Beijing's economic development zone, as long as the "radish run" is turned on, passengers may hit a self-driving car with no safety officers.
Baidu Apollo's first unmanned manned test license is not only a small step in the unmanned use of Baidu's autonomous driving, but also a big step in China's autonomous driving industry.
The success of Baidu Apollo's "moon landing" has set off a global autonomous driving industry competition.
01
The global autonomous driving storm sounded the "three-step" championship
Achieving "unmanned" autonomous driving is not simpler than landing on the moon.
As Zhang Junyi, managing partner of Oliver Partners, said to the media, "The development goal of autonomous driving is not to replace people, but to surpass people".
There are many difficulties in wanting to "surpass human" automatic driving, such as at this stage, due to the influence of policies and other factors, it is impossible to achieve real no one, and commercialization is difficult to land.
On the difficult road to the moon, many self-driving head players have emerged. According to Navigant Research's 2021 autonomous driving rankings, Waymo, Baidu, Cruise, and Argo ranked in the first echelon of global "Leaders", completely opening up the gap with the second echelon such as Mobileye.
The first echelon of global autonomous driving is roughly divided into two factions, Waymo, Cruise, Argo, and Baidu Apollo. Waymo pie is a U.S. company.
Among them, in 2019, Waymo made a name for itself with a valuation of $175 billion, becoming the world's most valuable self-driving company.
Although Baidu Apollo was established later than Waymo, it quickly rose on the world stage with the trend of small steps and fast running, and formed a two-strong pattern of world automatic driving with Waymo.
As of March 2022, Baidu has more than 1,000 patents for high-level autonomous driving in the world, ranking first in the world, and has the potential to surpass Waymo.
For global autonomous driving companies, in order to achieve the unmanned "moon landing" of autonomous driving, it is necessary to solve the two major problems of unmanned and commercialization, which can be disassembled into three steps of technology, product and operation, and Baidu's move beyond Waymo is hidden in it.
The first step of technology: In order to ensure the safety of automatic driving, the autonomous driving industry is pursuing technical redundancy. Compare Waymo and Baidu, and you'll see that they've taken two very different technical paths.
Waymo's technological development route is based on bicycle intelligence, using multi-line lidar perception.
Take the Waymo unmanned vehicle as an example, the body is equipped with 5 lidar and 4 77G millimeter wave radars, but the future of this saturated perception of bicycle intelligence is still questionable.
Li Jun, an academician of the Chinese Academy of Engineering, mentioned at the 2020 World Intelligent connected vehicle conference:
"Although Waymo's perceptual performance is very sensitive, its disadvantage is that it is very dependent on high-precision maps, the cost is very high, and bicycle intelligence does not make full use of the information and physical fusion method, which is difficult to solve the problem of 'unclear rules'."
Baidu Apollo takes a "car-road-cloud" collaborative route, does not rely entirely on the perception system, but uses bicycle intelligence, monitoring safety system and parallel driving to complement each other to form a "three-layer safety system", which can ensure the safety of vehicles and personnel to the greatest extent in any uncontrollable situation on one side.
In terms of the decision-making planning system of bicycle intelligence, automatic unmanned driving needs to quickly make intelligent decisions to the outside world within 100 milliseconds to ensure safety.
Baidu Apollo starts with traffic rules, and inputs a large number of abstracted traffic rules into the system with policy expectations.
Driving means risk, and the input of road laws and regulations is far from enough. In addition to entering traffic regulations, Baidu Apollo also adopts an AI learning method to determine the over-decision of obstacles through data accumulation and algorithm training.
It can be expected that after accumulating enough data, the unmanned vehicle may also be able to identify and even infer the next action of pedestrians in a very short period of time in the future, and complete the "super decision" in a very short period of time.
The second step of the product: from the perspective of product dimension, Baidu's unmanned vehicle cost is lower than Waymo, which is more conducive to forming a commercial closed loop.
First look at Waymo, because Waymo body is equipped with multiple Honeycomb lidar, the cost of unmanned vehicles has been high.
Former Waymo CEO John Krafcik revealed in an interview with Manager Magazine that if Waymo equipped jaguar I-Pace with sensors and computers, it would not cost more than the price of a mid-range Mercedes-Benz S-Class.
Based on this estimate, the current cost price of Waymo's front-loading mass production model with Jaguar models is about 180,000 US dollars (about 1.17 million yuan).
Obviously, Baidu has won the cost war of autonomous driving compared to Waymo.
In June last year, Baidu Apollo and Jihu cooperated to launch the shared unmanned vehicle Apollo Moon, based on the "ANP-Robotaxi" architecture, which makes the vehicle sensor suite more lightweight and the cost is greatly reduced, cutting the previous model by half.
The cost price of 480,000 is only equivalent to the price of L2+ high-end brand cars on the market.
With Baidu's in-depth cooperation with more OEMs, it can be expected that the day when The "unmanned vehicle" produced by Baidu and the domestic mid-range electric vehicle will be at the same price will arrive earlier than we think.
The third step of operation: operation is the goldstone of automatic driving, only through the verification of passengers and payment to take the lead in the practice, in order to complete the enterprise's self-hematopoiesis.
At present, Waymo and Baidu are both running into the operational stage, but the order volume of the two is far from the same.
First of all, according to a data, according to estimates, Waymo can reach 26,000 to 52,000 orders per quarter, while Baidu's "Radish Run" in the fourth quarter of 2021 has a manned operation order volume of 213,000 orders, which is 4 to 8 times the order volume of Waymo.
The disparity in order volume is actually the gap between the scale of the cities operated by Waymo and Baidu.
Baidu's Radish Run has achieved landing operation in 9 cities in China, including many super cities with a population of more than 10 million in Beijing, Shanghai, Guangzhou, and Shenzhen, while Waymo is currently only landing in Phoenix and San Francisco in the United States, while Phoenix's annual resident population is only one million, which limits the influence of autonomous driving.
However, Baidu Apollo took the lead in igniting the torch of autonomous driving in the center, making it spread to other cities faster.
From the perspective of technology, products and operations, compared with Waymo's perception redundancy, Baidu Apollo uses a three-layer security system to solve the underlying security problem of "unmanned vehicles", and the security is higher;
At the product and operational level, Baidu Apollo "unmanned vehicle" cost is lower than Waymo, but the order volume, operating cities are more than Waymo, compared to Waymo, Baidu Apollo can complete the "unmanned vehicle" commercial closed loop faster, the first to enter the global autonomous driving "no man's land".
As a result, Baidu Apollo overtaking Waymo has become a foregone conclusion.
02
Behind autonomous driving:
Policy, geography, and financing environment are concentrated
Autonomous driving is not only a corporate war, but also a centralized war related to policy, geography, and financing.
From a policy perspective, China and the United States maintain a positive and positive attitude towards automatic driving, but the pace of the US policy rhythm is bold and flexible.
Since 2016, the mainland has issued more than 10 national policies to ensure the all-round development of autonomous driving from the development of digital transportation and intelligent networked vehicles.
The notice of unmanned manned demonstration application issued this time is an important proof that Beijing has actively increased the intensity of policy first-mover testing under the background of the "Outline for the Construction of a Transportation Powerhouse", and it is also a major breakthrough in the development process of China's automatic driving.
Because U.S. states have different attitudes toward autonomous driving, there is also a gap in the degree of policy tightening.
For example, Waymo's california, where it is headquartered, is more relaxed, allowing self-driving cars to be tested on public roads in 2012.
In 2017, new regulations on autonomous driving were revised to allow the testing and deployment of fully autonomous vehicles, that is, unmanned vehicles have achieved true unmanned road testing. The densely populated state of New York was more restrictive, with tests only opened in 2017.
Although the policy of local autonomous driving development in the United States is relatively relaxed, the policy openness of US states for commercial operation is not as good as that of China.
Deng Zhidong, a long-term professor of the Department of Computer Science of Tsinghua University and director of the Visual Intelligence Research Center of the Institute of Artificial Intelligence of Tsinghua University, told the heart of the car that from a business point of view, Beijing and other places allow Baidu Apollo to carry out trial operation of self-driving taxis with safety officers and charge for trial operation, which is also a major breakthrough in policy.
At present, Radish Run has begun to commercialize paid operation services in Beijing, Chongqing, and Yangquan, while the United States has not yet allowed Waymo, Cruise and other companies to charge driverless orders, which means that in the Robotaxi business, these American autonomous driving companies can not get revenue from the C-end for the time being.
From the perspective of geographical environment, the participants and complexity of China's roads are high, and the road conditions and scenes in California and Arizona are relatively single.
The data shows that the density of traffic participants on Yizhuang roads in Beijing is 15 times that of California roads in the United States.
The mountain city of Chongqing is enough to confirm the diversity of mainland roads, Chongqing has many mountains, large overpasses, and sometimes even encounters super-large interchanges with very complex road conditions such as the Huangjuewan Bridge.
How complex is it? The bridge has 5 floors, each leading to a different direction, and 15 ramps, connecting important nodes such as the Chaotianmen Yangtze River Bridge, the Cimushan Tunnel, the Inner Ring Expressway, and the Airport Expressway.
In China, such complex roads abound. This means that the Apollo "unmanned vehicle" not only has to go to ordinary highways and urban expressways to test, but also to drive up the mountain, drive on the highway, and complex viaducts, but these also accumulate sufficient and diversified data experience for Conercase, and improve the safety of Apollo.
From the perspective of the financing environment, the fierce knockout race of unmanned autonomous driving has screened out a large number of American companies.
Before ZOOX sought to sell itself, after waymo's valuation continued to be reduced to $30 billion, in the past year, Google's parent company Alphabet was dissatisfied with waymo's slow progress in autonomous driving, and Waymo's CEO, CFO and other executives left.
The failure of these companies is related to profitability, and the autonomous driving business of American companies cannot be profitable for the time being due to taking the bicycle intelligent route.
However, Radish Run has opened test ride services in 9 cities, and has begun to commercialize paid operation services in Beijing, Chongqing and Yangquan, and the number of orders in 2022 is still growing at a high speed.
Obviously, Baidu Apollo's multi-legged walking profit model has brought more vitality to China's autonomous driving industry.
At present, autonomous driving companies are looking forward to "landing on the moon", but it is not enough to complete the moon landing.
In order to maintain the sustainable advantage of China's autonomous driving industry represented by Baidu Apollo. Professor Deng Zhidong, director of the Visual Intelligence Research Center of the Institute of Artificial Intelligence of Tsinghua University, believes that under the premise of ensuring absolute safety, relevant government departments can stimulate innovation through appropriate advanced policy planning and deployment.
"Policy breakthroughs with points and areas can not only make policy formulation keep pace with the rapid development of autonomous driving in the United States and Europe, but also be more conducive to the commercial landing of China's autonomous driving industry." Professor Deng Zhidong explained.
In fact, it is not unreasonable that the battle of autonomous driving is called the battle of the future. From a commercial point of view, it can bring a trillion-level market in the future, but at this stage it is also a gold-devouring beast, from the perspective of social value, autonomous driving unmanned can bring an unprecedented traffic change, but the premise must be to ensure absolute safety.
This is like the process of the nations' struggle for the poles described in "When the Stars Shine", "Their struggle is no longer just to seize the poles, but also to make their national flag fly first over the new lands." Autonomous driving is like that, betting on the present, fighting for the country and the future.
When will the day of the official confrontation between the autonomous driving of China and the United States come? We can't accurately predict it, but judging from the competition between Baidu Apollo and Waymo, Baidu Apollo has won the autonomous driving highland on behalf of Chinese autonomous driving companies. We will not be too far away from achieving "real nobody" in autonomous driving.