sail to the future｜autonomous driving is booming
2024-10-02
한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina
what can self-driving cars do? just 4 years ago, few people could fully answer this question. but today, the answer is "written" in the autonomous driving demonstration area in the southeast of beijing: unmanned retail vehicles, unmanned express vehicles, unmanned patrol vehicles, unmanned sweepers, self-driving passenger vehicles, self-driving freight heavy trucks... self-driving cars that once appeared in science fiction blockbusters have "driven" into life.
with the continuous iteration of the world's first "vehicle-road-cloud integration" autonomous driving demonstration zone, highway testing, unmanned demonstration operations and other policies continue to be "updated", and 600 square kilometers of roadside digital equipment and smart cities have been implemented with full coverage of private networks, all eight major application scenarios have been implemented in just 4 years. the dream of autonomous driving has become a reality, and a new industrial ecosystem is flourishing.
eight autonomous driving scenarios “dreams come true”
in the neolithic distribution center located in yizhuang, busy automated sorting lines are constantly loading express delivery onto delivery vehicles, which then carry the express delivery to thousands of households. the car is not big, but it can deliver hundreds of express items at a time. the entire process is completed by autonomous driving.
in beijing, hundreds of new unmanned vehicles are already on the road. just before 2020, this type of autonomous vehicle could not find a real application scenario, and it could not even be legally and compliantly put on the road. “the pain at that time was that we couldn’t find a place where self-driving cars could be tested.” when yu enyuan, the founder of neolithic huitong company, who was holding innovative results, heard that beijing was going to build an self-driving demonstration zone in yizhuang, he realized my own achievements have a "way out". on the second day of his inspection in yizhuang, he moved the company here.
in september 2020, the beijing autonomous driving demonstration zone was officially established, and the company became one of the companies participating in the formulation of management standards and technical standards for the demonstration zone; more than half a year later, it became one of the first companies approved to test on roads in beijing. obtained the first road permit for unmanned delivery vehicles.
"we established the standard in yizhuang, and then replicated and promoted it across the country." yu enyuan said that in just 4 years, this cute-looking unmanned vehicle is "driving" from beijing to the country and the world, and it has been used in 13 countries. it has been implemented in more than 50 countries and cities, and the cumulative mileage of autonomous driving exceeds 10 million kilometers.
the emerging future industry is a new thing. there are many gaps in policies and regulations, and there is an urgent need for breakthroughs and innovations. it is precisely in order to explore innovative regulatory measures for the application and promotion of new intelligent network technologies, new products, and new models, and to increase the intensity of policy implementation and pilot testing. in april 2021, this city established the first intelligent network-connected vehicle in china relying on the demonstration zone. policy priority area.
also taking the lead in achieving unmanned breakthroughs in self-driving passenger cars in yizhuang is pony.ai. "since setting up a domestic r&d center in beijing in 2017, we have accumulated more than 31 million kilometers of autonomous driving miles in various places, which is equivalent to a lifetime of driving experience for dozens of ordinary drivers." said the relevant person in charge of the company.
"going first" means stimulating technological innovation and industrial innovation through policy innovation. the relevant person in charge of the municipal autonomous driving office said that in more than three years, the city has taken the lead in the country in releasing many innovations such as morning and evening peak testing, mutual recognition of off-site testing, unmanned delivery vehicles on the road, highway testing, and unmanned demonstration operations. the policy has enabled the coordinated development of eight major categories of urban application scenarios, including autonomous passenger vehicles, unmanned delivery vehicles, and unmanned retail vehicles.
with policies and application scenarios in place, the "gravitational field" for industrial development will naturally be formed. as of now, there are 32 test car companies in the demonstration area, with a cumulative autonomous driving mileage of more than 30 million kilometers. "beijing is accelerating the technological innovation and industrialization of intelligent connected vehicles. all leading companies are here, more than a hundred companies are participating in the construction of demonstration zones, and the number of operating vehicles leads the country." the relevant person in charge of the municipal economic and information bureau disclosed that in the first half of 2024 the output value of beijing's intelligent connected new energy vehicle industry exceeded 226 billion yuan, a year-on-year increase of 16.1%, and the transformation and upgrading has achieved remarkable results.
the implementation of the “china plan” for autonomous driving
just as yu enyuan was preparing to put autonomous vehicles on the road, xuan zhiyuan, chairman of yukong zhixing technology company, was running for the implementation of the "chinese plan" for autonomous driving - "vehicle-road-cloud integration."
in layman's terms, "vehicle-road-cloud integration" means the cooperation between smart cars, smart roads, and powerful clouds to make cars run safer and more efficiently. as early as 2016, my country launched the "vehicle-road-cloud integration" technology roadmap, but there has been no city-level scenario implemented.
a turning point came in september 2020, when beijing took the first step and launched the construction of the autonomous driving demonstration area, which is the world's first "vehicle-road-cloud integration" high-level autonomous driving demonstration area. at road intersections in yizhuang, in addition to traffic lights, street lights and other traffic facilities, light poles are also equipped with sensors such as millimeter-wave radars and cameras. data such as traffic lights and congestion at each intersection can be grasped in real time and can be aggregated to the cloud control basic platform.
xuan zhiyuan is responsible for building a basic cloud control platform for the demonstration zone: "this platform is like the 'eye of god', which can help autonomous vehicles improve their collaborative sensing capabilities and provide collaborative decision-making and planning suggestions."
the "green wave access" that yizhuang has achieved is precisely the cloud control basic platform that plays a role behind the scenes. on the one hand, the cloud control basic platform accesses traffic control data and optimizes the length of traffic lights to make the green lights in directions with heavy traffic last longer to avoid congestion; on the other hand, it can plan the global optimal driving route for self-driving cars and calculate safe under the premise of driving at what speed can you drive smoothly all the way.
every day, the cloud control platform can receive and process about 300t of data, and has achieved data access for more than 420 intersections and more than 800 self-driving vehicles in the beijing economic and technological development zone. as a data base, the cloud control platform also supports 67 roads in the economic development zone to achieve green wave traffic and 257 intersections to achieve dynamic optimization. in the past year, the average vehicle delay at intersections decreased by 33%, and the average vehicle speed increased by 45%.
"not only that, the demonstration area promotes the integrated development of 'vehicle-road cloud network map', fully verifying the 'vehicle-road-cloud integration' autonomous driving technology route." the relevant person in charge of the municipal self-driving office introduced that the car terminal was the first to allow passenger use vehicles are fully unmanned for testing, and unmanned delivery vehicles are on the road. so far, more than 800 test licenses of various types have been issued; at the road end, the construction and model of the standard intersection of "multi-poles in one, multi-sensors in one, and multi-boxes in one" have been continuously optimized. promotion, the construction cost has been successfully reduced by more than 50% compared with the 1.0 stage; on the cloud, the city's unified cloud control platform will be built, which will be built into the world's first digital management platform with complete functions, advanced technology, and the largest amount of uploaded data in the future; on the network side, ultra-high-speed intelligence will be built the urban private network has become the first in the country to connect passenger cars and unmanned delivery vehicles. remote driving communication link; map end, completing the collection and production of high-precision maps of 160 square kilometers within the economic development zone.
“gravity field” gathers new industrial ecology
the rise of an industry requires both accumulation and success, as well as running through the commercialization cycle. for self-driving cars, cost reduction is the threshold that must be crossed to achieve commercialization. among the many parts and components, the sensors that allow the car to see the road clearly have always been the most costly.
"if you want self-driving cars to perceive the world like humans, then a perception terminal like the human eye and a brain-like processing system are essential." ceo of the intelligent robot company, a startup company born in tsinghua science and technology park three years ago following this idea, shan yi led his team to develop an "ai binocular camera" that uses two cameras to replace lidar. the ai binocular camera sees more, and has a panoramic view of three-dimensional information that is 10 times denser than lidar.
in addition to brighter eyes, there are also smarter algorithms. fish-eye cameras and side-view cameras are generally placed on both sides of mass-produced cars with intelligent driving capabilities. the former is wide-angle and is used when parking, while the latter is used while driving. "can the fisheye camera be used while parking and driving at the same time?" the fisheye smart driving solution proposed by jianzhi robot company realizes the reuse of the surround-view fisheye camera while parking.
after three years of research, this start-up company with numerous technologies has reached a critical moment of industrialization. this year, shan yi moved the company into the demonstration zone and immediately received pre-b round of financing jointly led by beijing economic development zone industrial upgrading fund and beijing intelligent connected automobile industry fund, with the financing amount reaching 30 million us dollars. "this is the area we have come across that knows the most about autonomous driving. the intelligent connected car industry ecology is very complete, which will help us find upstream and downstream partners and complete the puzzle of the commercialization strategy." he said.
from hardware such as cameras and chips, to software such as algorithms and systems, to intelligent connected cars, an industrial chain has gradually gathered and developed. data shows that the demonstration zone has gathered more than 120 companies in the field of intelligent connected cars, including more than 50 high-tech companies, 32 specialized new companies, 9 small giants and many unicorn companies. "beijing will focus on intelligent network connectivity, continue to lead the development of the industry, promote the opening of key scenes such as airports and train stations, improve the legal and regulatory system, promote the resolution of common and difficult issues in the development of the autonomous vehicle industry, and help the intelligent network vehicle industry to achieve high-end quality development." said the relevant person in charge of the municipal economic and information bureau.
expert comments
yuan yu, deputy director of the national intelligent and connected vehicle innovation center:
autonomous driving promotes the implementation of various innovative technologies
as a high-level autonomous driving, l4 level autonomous driving means that the vehicle can autonomously complete the driving task from the starting point to the end point under specific conditions and environments, without driver intervention. yuan yu, deputy director of the national intelligent and connected vehicle innovation center, said that this is not only a technological leap, but also can bring changes to transportation methods as the technology continues to mature, allowing passengers to experience a higher level of traffic safety and more safety. good urban traffic efficiency.
yuan yu believes that from the perspective of industrial development, autonomous driving is a representative of "new productivity". an autonomous vehicle integrates various high-tech technologies such as big data, large models, artificial intelligence, cloud computing, and 5g communications. autonomous driving drives the implementation of these high-tech technologies, becoming the "mother ecosystem" of these technology application ecosystems and promoting a virtuous circle of related industries.
in recent years, my country's autonomous driving technology has continued to make breakthroughs, and the "vehicle-road-cloud integration" technical route is also accelerating its implementation. yuan yu emphasized that if you want a self-driving car to truly become a skilled "experienced driver", you need to continue iterative upgrades in sensing, decision-making, control and other technologies. among them, in terms of environmental perception, in addition to solving the problem of lossless transmission of sensory data and reducing the risk of data recognition errors to a certain extent through "end-to-end" technology, it is also necessary to improve the relationship between the vehicle and the roadside in the "vehicle-road-cloud integration" environment. fusion sensing capabilities of the device. at the same time, autonomous driving requires very fast response, and the vehicle's decision-making and response need to reach the millisecond level. therefore, equipment technology such as in-vehicle smart driving chips should also be further innovated to make the vehicle's judgment of the environment more like an "experienced driver" rather than a frustrated machine.
yuan yu believes that autonomous driving is still a new thing, and its industrialization and commercialization require not only testing and improvement in real environments, but also rapid growth with social tolerance; it also requires continuous improvement of policies and regulations, such as the current "vehicle-road-cloud integration" relevant standards and regulations still need to be clarified in the construction of a "humanized" environment. after autonomous vehicles are put on the road, there is an urgent need to fill the regulatory gaps in liability identification and handling.
with the continuous upgrading of technology, autonomous driving is expected to reach a higher level in the future and achieve l5 level. yuan yu said that l5 represents a fully automated driving system, which can even be imagined as "ultimate driverless driving." the vehicle can drive automatically in any environment, any road conditions, and any weather conditions, without the need for manual intervention at all.
super chain
as a typical application of artificial intelligence, self-driving cars perform driving tasks by machine intelligence. my country's "vehicle driving automation classification" divides autonomous driving into levels l0 to l5. l3 is the dividing line. the following is assisted driving and the above is autonomous driving. l4 means that in most scenarios, the vehicle can achieve autonomous driving, but human intervention is required in extreme cases; l5 can be achieved in any scenario. autonomous driving. the industry believes that l4 and above are high-level autonomous driving, which can achieve "hands-off", "eyes-off" and "feet-off" during driving.
