The extreme end-to-end AI smart car conference with the theme of “evolving AI reshaping smart car” was held in Chengdu on Aug.
28, and the new ASD V2.
0 software was officially released and will be launched on Ji Yue 07.
ASD V2.
0 software brings the most important update since the launch of the Jiyue model, covering more than a dozen major modules, such as end-to-end large model smart driving, large model intelligence module, intelligent car couplet, and so on, totaling 947 new features and optimization upgrades (you can click to view the video introduction).
ASD V2.
0 is expected to be officially launched by OTA on September 10th.
, “Ji Yue has become the first smart car brand in the world to apply the L4 self-driving model Baidu Apollo ADFM,” said Xia Yiping, CEO of Ji Yue.
“this is the first time that L4 self-driving technology has enabled a household car! It is the world’s first large self-driving model that supports fully self-driving applications-the deep integration of Baidu ADFM and Jiyue.
Baidu has more than a decade of continuous research and development in the field of AI and self-driving, and more than 150 billion of the continuous investment is the key to the more sustainable leading force.
Without 50 billion, you can’t do smart driving, and smart cars that cost billions are all ‘road killers’.
” , 
, PPA comprehensive upgrade ASD extremely open-book pure visual end-to-end, PPA smart driving upgraded to ASD (Apollo Self Driving) Apollo high-level intelligent driving, marking the more pure visual high-order intelligent driving into the end-to-end large model era.
Xia Yiping said that compared with the classical smart car with “lidar + high-precision map + rule algorithm”, the new smart car with “pure vision + end-to-end large model” has opened a new stage in the development of intelligent driving.
With the rapid evolution of the AI model, the role of “crutches” of lidar and high-precision maps will be gradually replaced, and pure vision will continue to improve the level of intelligent driving with its strong self-learning ability.
, , Xia Yiping said that Baidu is the world’s leading AI model company and the leader in self-driving technology, with more than 10 years of accumulation in the field of AI technology and self-driving, and the world’s leading technical capability.
At present, Apollo’s L4 self-driving mileage has exceeded 100m km, and has established a super-5.
5EFLOPS high-computing training cluster, which can achieve efficient training and fast iteration of the model.
Based on this, the more it will achieve “monthly evolution” and provide more and more advanced, intelligent and safe intelligent driving experience.
The perception ability of extremely pure visual high-order intelligent driving is even higher than that of the mainstream lidar, its 3D accuracy can reach centimeter level, the surrounding environment of the car body is fully covered by 360 degrees, the sensing distance is up to 250m, and it can identify mineral water bottles within 200m.
The identification accuracy of the speed of moving obstacles is also controlled within the error range of 0.
1 m / s, and it can also distinguish green plants, temporary construction fences, fault vehicles parked on the side, and predict pedestrian trajectories.
Even in the rain and snow haze, or in the night, tunnels and other lack of light environment, it can still be accurately perceived and identified.
is still the “experience king” at a time when the smart car has reached a new height.
The ASD of Jiyue will bring users a silky and smooth smart driving experience, combined with automatic gear shift, automatic steering lights and other functions, so that users can get on the car and go, making it easier to drive, such as “buy the more you get to the AI driver”.
ASD includes navigation assistance, valet parking, high-end lane maintenance and active safety, with a total of 106 new and upgraded functions, realizing the overall evolution of driving, safety and parking capabilities.
In terms of driving ability, ASD can achieve free movement from point A to point B, combine lane change and avoidance into one, intelligent speed limit, automatic traffic light start and stop, autonomous bypass, continuous lane change, courtesy of pedestrians, unprotected left turn and other functions to comprehensively improve the intelligent driving ability and easily cope with various complex scenarios on high-speed viaducts and urban roads.
In particular, it has a strong ability to avoid obstacles, accurately grasp the opportunity of avoidance, and make more human-like decisions.
at the same time, using AI computing power to accelerate modeling, it can detect 5000 collision risks per millisecond and calculate 5000 distances, and actively explore the best driving path.
The ability to pass through the intersection efficiently, so that the car can learn to take the initiative to detour when it encounters traffic in front of it.
While changing lanes to avoid high-risk vehicles, ASD will pay extra attention to open a safe distance to avoid concessions to enhance the user’s sense of peace of mind.
By the end of 2024, ASD will be available in places with Baidu navigation maps, so that it can be opened all over the country.
, 
, in terms of the evolution of security capabilities, the forward AEB capability has been further improved, expanding the effective scope of action, helping users to decelerate to the maximum extent in a timely manner, striving for more safety time and reducing the degree of harm.
Add two-stage braking, optimize the AEB algorithm strategy, first decelerate and then brake vigorously, further improve the safety and experience of smart driving.
The new bend AEB function can accurately deal with the sudden crossing behaviors of pedestrians, express cars and takeout cars in the corner, and help users to stop in time.
, 
, parking capacity also ushered in a substantial upgrade.
APA automatic parking assist not only supports many kinds of complex parking spaces, such as vertical parking spaces, horizontal parking spaces, space parking spaces, reverse parking spaces, broken road parking spaces, etc.
, but also solves the parking problem of dim light and extremely narrow parking spaces, and the parking success rate is doubled.
In addition, parking will also automatically put away the rearview mirror to prevent rubbing, giving play to the hard-core parking technology comparable to the “old driver”.
, the big model brain module has once again evolved the industry’s first lip movement recognition.
As early as October 2023, SIMO took the lead in merging Baidu Wenxin to become the world’s first voice assistant to apply the AI model, which can be called the “ceiling” of voice interaction in the car world.
The recognition speed of SIMO is not only as fast as 500ms, but also can interact smoothly even in weak / non-netted environment, supporting almost all functions of the whole vehicle, even solving math problems and making travel strategies are easy, just like an “omnipotent butler”, bringing a more human-like, more natural and smooth human-computer interaction experience.
, 
, at this conference, Ji Yue also released the industry’s first native AI large model cockpit.
Based on the new native AI large model architecture, the underlying capabilities of all cockpits are reconstructed, and a set of end-to-end cloud collaborative AI agent architecture is created, which models the perception, understanding, execution and other capabilities of the cockpit, which not only retains the end-to-side rapid response capability, but also integrates the strong understanding, generation and decision-making capabilities of cloud-based large models.
, 
, , the newly upgraded SIMO 2.
0 is smarter and more humane, subverting the traditional question-and-answer rigid interaction of voice assistants.
SIMO 2.
0 has many capabilities, such as interesting exploration, role-playing, scene mode, and so on.
For example, when driving, you can ask about the housing price in the neighborhood near SIMO and recommend one.
It is not a problem to play the role of a tutor, a car expert, and a raiders expert.
It also supports 10000+ user-defined scene settings.
, SIMO 2.
0 integrating multimodal large models realizes the industry’s only lip movement recognition function.
By integrating and recognizing the user’s lip movement trajectory, the speech recognition accuracy in noisy environments is increased to 90%.
At the same time, SIMO 2.
0 also upgrades dialect recognition capabilities and supports multiple dialect dialogues.
With the support of multiple command simultaneous recognition, multi-tone zone Group chats, and full-scene voice commands inside/outside the car, it creates a very futuristic immersive smart cockpit for users.
,, and HUD are transitional products serving “people’s driving” and have not been adopted on Geo vehicles.
The design of Geo smart cockpit fully considers the popularization of high-end smart driving and even future unmanned driving.
Human-computer interaction experience.
Increasingly innovatively integrates lane-level navigation maps with smart driving SR information.
By completely reproducing the physical world, it enriches elements such as green plants/water systems, refined lane lines/zebra crossings/traffic lights, and even front and rear lights.
1:1 Presented on the large screen, giving users an immersive navigation experience, making it the “next generation of high-definition AR HUD.
” The car map has also been upgraded to Baidu Map V20, realizing “mobile locomotive, navigation as one”.
The world’s only 3D lane-level navigation covers cities across the country, making it truly available across the country.
, 
,,UI design renewal is also a highlight of this smart cabin upgrade.
A new future theme “Creation of the Era” has been added to the extremely exciting fully immersive leisure cabin, and a more user-friendly UI design The improvement brought by the user experience is self-evident.
In terms of entertainment ecosystem services and smart interconnection, the V2.
0 version launches Tencent START cloud game, which allows you to play dozens of popular games such as Black Myths: Wukong and Original God with one click.
The cockpit can also be connected with small ecology, smartphone/watch wearable devices, switch projection to achieve interconnection.
The fully revitalized App 3.
0 makes it more convenient to prepare and find cars with one click, bringing users a cutting-edge people-car and home interconnection experience.
As a Geyue brand that has won three consecutive championships in the 2nd China Smart Driving Competition, how does the upgraded ASD V2.
0 version perform? Everyone is welcome to pay attention to the Guangzhou Station of the 2nd China Smart Driving Competition on September 22.
We look forward to the extraordinary performance together.
, Return to First Electric Network Home>.
