It has always been human nature to free your hands. Since 1886, when German Karl Benz succeeded in the world's first three-wheeled car, people's imagination of autonomic driving has never stopped.
In 1969, concepts such as "the will of the car owner" and "the car can drive by itself" appeared in the movie "The Almighty Golden Turtle Car". In 1994, the Explorer driverless car also appeared in the movie "Jurassic Park". In 2000, the intelligent car KITT in the movie "Thunderbolt Ranger" changed color through automatic transformation, which made the black technology of automatic driving show a specific imagination.
Nowadays, intelligence has completely swept the automotive industry, and technologies related to autonomous driving are also constantly landing. From Tesla to a number of new forces, there are layouts and technologies in automatic driving.
However, at the 3rd Changan Automobile Science and Technology Ecology Conference, when the new automobile Changan Qiyuan CD701 was like a "model robot", the demonstration of automatic warehousing, automatic charging pile charging, automatic gun pulling and other "self-reliant" functions, still made the outside world open their eyes and exclaimed. The arrival of Changan Qiyuan CD701 once again pulled the car intelligence to a new height, so that the imagination of the car autonomous driving in the movie has achieved partial mass production.
Because Changan Qiyuan CD701 has an ability - self-learning, it has the ability to self-evolve based on the integration of the end cloud. In other words, a traditional car is a functional machine, while the Kaizen CD701 is an intelligent machine that can learn evolution. "Car end" and "cloud" complement each other in the forward and reverse of data and computing power, combined with OTA technology, to achieve continuous iteration of functions and continuous optimization of experience, so that the car can be driven smarter and easier to use.
For example, when we commute to work, CD701 can learn our traffic environment, commuting route, driving habits and emotions, etc., automatically help us adjust driving mode, driving route, air conditioning temperature, parking location, and automatically evolve memory driving and memory parking services. It can even look like the scenario demonstrated at the press conference: when the people in the car get out of the car to see the sunrise, drive themselves and look for smart charging piles.
Of course, if you want to become a smart machine that learns to evolve, there must be strong hardware + software support behind it. Changan Automobile has built the SDA platform architecture, and at the same time, through software, the new car CD701 has three capabilities: "intelligent brain", "intelligent body" and "intelligent service". Behind any capability, sufficient professional R&D personnel, powerful computing power, huge data collection, scene collection, and so on are required.
Take the "CD701 commuter route self-learning evolution" as an example, the car needs to be able to learn the user's commuter route from home to work, and then optimize the driving mode, driving route, air conditioning temperature and so on. Behind the ability of CD701 to learn commuting routes, there is a self-learning pilot system that automatically learns the map, which requires Chang'an's in-depth opening in the map.
At present, the entire autonomous driving team R&D project in Changan Automobile is divided into planning, control, decision-making, perception, positioning, and maps, of which about 70 people are responsible for the research and development of maps.
Some people may ask, there are very top map development teams on the market, and even more R&D personnel than Changan Automobile, why should Changan Automobile's map development team be emphasized here? Yan Kai, the chief engineer of Changan Automobile's map development, answered this question with his own example.
Yan Kai joined Changan Automobile in early April last year, and after 17 years of working experience in a picture manufacturer, the biggest driving force for him to come to Changan Automobile is that "maps can be close to applications, close to data collection applications". Yan Kai said, "The best way to produce autonomous driving is for car companies to do this." ”
Compared with map vendors, Changan Automobile puts map research and development under the autonomous driving section, which can allow map developers to deeply understand how to better apply maps in the self-driving system on the car.
The development of technology based on mass production and application is Changan Automobile's thinking in the field of autonomous driving. In Yan Kai's words, Changan Automobile's autonomous driving will not pursue unrealistic goals, nor does it want to be the "first" in the industry. Instead, it can be practically applied in the car, applied in places where users are interested, and truly use intelligent technology to bring changes to users.
In the author's opinion, if many leading technologies released at the Changan Ecological Conference, such as the "UNIBrain Super Brain Central Computing Platform" with a computing power of more than 1000Tops, and the BEV perception algorithm that can complete a minimum of 1500 perceptions per second, are Changan Automobile's strength in the era of digital intelligent vehicles, then being able to apply and mass produce, that is, pragmatic, is the original intention of Changan Automobile research and development technology.
On the other hand, the Changan Automobile Map R&D department is located in the Changan Technology Company Park, which was just launched at the end of last year. This park targets more than 3,000 people, specializing in new automobile research and development, and the Changan Global R&D Center not far away has an essential difference in mission and vision.
This is also the embodiment of the underlying logic of Changan Automobile's digital intelligence transformation and the creation of new cars: the research and development of technology is not to add to the original system, and the new car does not add some functions to the traditional car, but completely breaks and redefines the logic of car building with digitalization, so that the car can achieve self-learning and evolution.
What Yan Kai and his map development team are currently doing is how to make the car learn more scenarios by itself with less data; How to reduce and simplify high-precision maps step by step, that is, to create a "heavy perception, light map" model, reduce dependence on maps, and serve the development of autonomous driving.
Yan Kai introduced the current advantages of Changan Automobile in the research and development of maps, on the one hand, there is no general definition of light maps in the industry, and each car company defines differently, and Changan Automobile's own research and development of maps can be defined through the traction of applications, in line with the rhythm of Changan Light Map's gradual implementation and release, and more in line with Changan Automobile's autonomous driving development.
On the other hand, it can make full use of the data of Changan Automobile's mass production vehicles for map development, that is, it has been automated when developing maps, and eventually achieve 100% automation of the entire map production process. Yan Kai explained that Changan Automobile's mass production car generates data, which can immediately transmit the data back, and produce the map scene without any human intervention.
In fact, people who study technology have a rigorous attitude, a habit of speaking with data, and obsessed with the field they study. It is precisely because of this that when Yan Kai talks to us, there are a lot of professional terms. In Yan Kai's eyes, the new car is also a data collector, providing more material for them to solve problems in the scene.
Isn't this the definition of a new car, not just a car but an open car ecology. For users, new cars are "hardware pluggable, scene can be orchestrated, ecology can be on-demand, and the system is self-evolving", in turn, for these data-obsessed R & D personnel, they are also exploring technical fields that can better serve human beings and are worth landing through new cars.
In fact, even more moving than data is the spirit of engineers focusing on research and development. It is not difficult to hear from Yan Kai's description that Changan Automobile engineers have the desire to "look up at the starry sky" and are more down-to-earth cultivation. It is the tenacity to endure loneliness that makes this complex digital intelligent car easier and simpler.