In modern society, a car is often entrusted with many social attributes beyond the function of travel. For example, it represents wealth, class, personal style, is symbolized as a "family member", or is portrayed as a private space under the pressure of life and work for a short time...
But before the process of automobile intelligence began, the car was always a passive" existence that required human drivers to control and drive. All people's good wishes about cars can only exist in the imagination of "car robots" in science fiction movies or cartoons.
How can a car become a "robot" that can communicate with humans, have a certain degree of autonomy, and can evolve itself, and truly become a "companion" existence of humans? The answer to this question is no longer unrealistic or far-fetched.
Car robots are already at hand. The key conditions to achieve this goal are the emergence of a new round of AI technologies represented by machine learning and deep learning, the maturity of autonomous driving software and hardware triggered on this basis, and the application of a large number of AI interaction technologies to cars through a "software-defined" approach.
Time, place, people. Under the trend of "human-machine symbiosis", Jidu Company stands at the intersection of the technology industry with great changes in the automotive industry and the landing of artificial intelligence applications, and takes the lead in the industry to propose the goal of building the world's first mass-produced "automotive robot", providing a "sample" significance for the automotive industry to build a future generation of intelligent cars.
So, how to create an "automotive robot" that is "oriented to mass production"?
We can explore this answer from the dual dimensions of car-building logic and car-building progress since the establishment of Jidu.
The logic of car building, its first question is: how to define a "car robot".
The concept of automotive robots comes from an idea proposed by Baidu founder Robin Li at last year's World Artificial Intelligence Conference. "The future of smart cars is more like intelligent robots, or conversely, the mainstream direction of future robots is more like a smart car." This can be seen as the starting point for thinking about this new species.
And what is the difference between a car robot and a car? At the Baidu Create conference at the end of last year, Robin Li once again gave a complete definition of "car robot".
First, the car robot has the intelligent driving ability of L4 level and can move freely; Second, the accurate recognition of human-vehicle interaction and speech semantics allows cars and people to "communicate naturally", understand user emotions, and respond to user needs in real time, which is a caring intelligent partner; Third, based on artificial intelligence, automotive robots can learn and iterate on their own, continuously learn the habits of user use and interaction, and continuously optimize various experiences of automatic driving and intelligent cockpits to achieve "self-growth".
"Free movement, natural communication, and self-growth" have become the three major product concepts that run through Jidu to create automotive robots.
Jidu's "automotive robot" has the characteristics of electrification and intelligent networking of newly built cars on the one hand, and the AI capability of robots with "free movement, natural communication, and independent learning" on the other hand.
Mobile, interactive and OTA, now some smart cars can also do it, so what is the difference between it and Jidu's automotive robots? In essence, there is a difference in "autonomous" ability between the two, that is, the difference between active and passive.
Specifically, what are the differences between automotive robots and ordinary cars.
Move freely.
The "free movement" of Jidu automotive robot means that it has L4 level automatic driving capabilities from the time of product launch, which can truly liberate the driver's hands. Theoretically, the driver only needs to tell the destination, and the car robot can deliver you through autonomous control.
Natural communication.
Many cars today are equipped with voice assistants, even if they can support continuous dialogue. But it is still far from the natural state of communication between people and people, and this basic question-and-answer is still a command-based operation.
For the "natural communication" of Jidu automobile robots, it is first necessary to have accurate speech recognition capabilities and semantic understanding capabilities, so that the car can understand the user's natural language, identify the real intention, gain insight into the needs of the user, and make the human-computer interaction level infinitely close to human-to-human communication. For example, when you answer the phone in the car and talk about going to a certain place to go to an appointment, the car robot with natural communication ability will actively ask the passenger whether they need to go to a certain place after hearing the relevant information, and the general car still needs the driver to actively issue instructions to switch navigation. This is the difference between active and passive.
Third, self-growth.
OTA is an essential technology for the continuous upgrading of future cars under the current "software-defined car". For the current models on the market that can achieve OTA functions, OTA upgrades depend on the functional updates pushed or optimized by manufacturers, which is still a passive "learning". For the "self-growth" of Jidu automotive robots, it is based on AI, big data-driven self-evolution, which can carry out self-learning and iteration according to the user's usage habits, continuously optimize automatic driving and intelligent cockpit experience, and achieve active learning and evolution.
The evolution from "car" to "person" is the evolutionary logic demonstrated by the product concept of Jidu automotive robot. Based on this understanding, you can see the changes that automotive robots can bring as a new species. Car robots become your chauffeurs, your job assistants, your trip companions, your AI "confidants"... Anything is possible in the future.
Easier said than done. Will the "imagination" of automotive robots be just the "water moon" that these technology companies show to the outside world. Obviously, judging from the one-year announcement of jidu's car building, they are really moving.
In January, the official announcement of the car, in March registered and established; In June, the "clay model" was exposed, and the modeling design was completed in July, and although the automotive robot of Jidu was not yet available, it already had its original "appearance".
At the beginning of June, Jidu announced the development of the L4-level electronic and electrical architecture JET to create a "nervous system" for automotive robots, and in July, Jidu has also entered the engineering development stage synchronously, and within 207 days of establishment, Jidu proposed SIMUCar (Software Integration Simulation Prototype), entered the pre-development stage of intelligent driving and intelligent cockpit, and began to create an intelligent "soul" for automotive robots.
Recently, Jidu Automotive Robot has successfully realized the intelligent driving integration of urban and high-speed dual domains in the SIMUCar road test, providing reliable verification for the ability of "free movement".
Jidu accelerates the creation of automotive robots, and naturally needs to create a strong industrial supply chain. Judging from the information that has been disclosed, Jidu has reached a cooperation with the auto parts giant Continental Group; Reached a strategic cooperation with Hesai Technology on lidar; In terms of smart cockpit, Jidu and Qualcomm have reached a cooperation to launch the 4th generation Snapdragon digital cockpit platform 8295 in China; In terms of intelligent driving, Jidu will use Baidu's latest generation of autonomous driving platform, while equipped with NVIDIA's strongest computing power Orin SoC intelligent driving chip.
The soul of the robot is being "forged", and the robot's vision, heart, brain center, and nervous system have also been "fixed", and the speed of concentrating on creating an automotive robot is really amazing.
Approaching the new stage of Jidu car manufacturing, we may be curious: what should Jidu's automotive robot look like?
What kind of unique vision system should automotive robots have? Jidu just gave his own thoughts.
Inspired by automotive robots, Jidu combined the concept of "pixel" with the letter "J" to create the industry's first automotive robot brand logo "Pixel-J". Pixels represent the basic elements of machine vision, from which the car robot can foresee all scenarios.
AI found the "color of AI" from countless kinds of physical experimental data - AI purple as the accent color of the set degree car robot.
Starting from the pixel concept of the core of the car robot, Jidu integrates the brand's young and avant-garde spiritual genes, and also creates a set of ROBO Font automotive robot exclusive fonts, as a unique text for automotive robots to communicate with humans.
At the same time, Jidu also disclosed the product design style of the automotive robot concept car for the first time, and it was revealed that the appearance of the mass-produced automotive robot will be 90% consistent with the concept car.
As a result, the individual characteristics and real appearance of automotive robots have been presented to the public.
The automobile has changed from "thing" to "person", which is the key connotation of the "change" encountered by the automobile industry. Jidu took the lead in creating an "automotive robot" not only for the automotive industry to create a new product category, but more importantly, for the industry to provide a new paradigm of car manufacturing, such as the innovative proposed SIMUCar, which can carry out intelligent pre-development of completely decoupling software and hardware, such as the mass production development of L4 automatic driving capabilities from the beginning, and raise the next 2-3 years of automotive industry competition to a higher competitive water level, creating a benchmark for intelligent cars.
No intelligence, no car building, is already the consensus of the automotive industry. In 2021, when many players are pouring into "building new cars", the "car robot" that Jidu wants to build undoubtedly represents the real evolution direction of the future of the car. It is conceivable that no matter the large number of new brands and new models that emerge next, although they are not necessarily labeled as "automotive robots", they will certainly evolve to the concept of "free movement, natural communication, and self-growth" covered by automotive robots.
Ten years later, large-scale car robots with advanced forms and multi-purposes have filled the scenes of our daily lives, and we have long been accustomed to being in the world of human-machine symbiosis that only appears in science fiction works.
However, people will always look back, going back to the past 2021, Jidu is spending effort to create this "car robot".