It's Wednesday again, and the official spoiler for the Ideal L9 arrives as promised.
If the theme of the last spoiler was "self-research", then from the content released in this issue - the ideal AD Max for intelligent driving systems, the key word is "safety".
Of course, you can also understand that the ideal L9 as a "super daddy car", safety is definitely the most core selling point.
As the ideal CEO Li Xiang said: Autonomous driving is the life of the ideal car. The ideal goal is to get a ticket to a self-driving track in 2025.
The ideal L9, or more precisely, AD Max, is a key node for Lee to further get tickets to autonomous driving.
Just as the iPhone 13 Pro is to the iPhone 13 Pro Max, the ideal intelligent driving system is named "Apple".
From the subtle differences expressed in these words, it can be seen that on the ideal L9, the ideal has made a comprehensive upgrade to this intelligent driving system.
Known as a "fully self-developed intelligent driving system", AD Max certainly has highlights.
01
What is the origin of a lidar with 128 lasers?
Keyword one: lidar of 128 lasers.
Judging from wednesday's spoiler lidar information, the ideal L9 specifically mentions that "forward-facing vehicle-grade lidar is equipped with 128 lasers."
Compared with the Tutton Falcon lidar equipped with THE NIO ET7, there is only 1 laser, and the Sagitar Juchuang M1 equipped with the Xiaopeng G9 has only 5 lasers. What's so special about this 128 laser?
For example, the number of lasers is to lidar, just like the cylinder is to the engine. The more cylinders, the higher the system performance and the smoother the operation.
The benefits are twofold:
First, the greater the number of lasers, the higher the resolution of lidar.
The number of lasers determines the number of laser beams that can emit light at the same time, and increasing the number of lasers allows the system to achieve higher resolution and more accurate perception of details of distant objects. Taking the semi-solid-state one-dimensional mirror scheme as an example, as many lasers as there are inside the lidar, how many lines can be achieved.
The Hesai AT128 equipped with the Ideal L9 has 128 lasers inside, which can simultaneously emit 128 lasers in the vertical direction to scan, and hit 1.53 million points per second in three-dimensional space (that is, the number of point clouds reaches 1.53 million per second), thus gaining a good perception of the surrounding environment.
In scenes where cameras are limited such as low light and strong light, lidar can ensure the accuracy of the system's perception of the environment, and help vehicles improve the recognition of stationary and special-shaped obstacles, such as accident stationary vehicles on highways, construction barricades and other objects, greatly improving safety.
Second, there are more lasers and also provide higher safety redundancy.
With only one or a few lasers, once the laser fails, it is very catastrophic for the perceptual data output of lidar, facing the risk of having no data at all.
With 128 lasers, even if one laser fails, the impact on the overall perception is very limited, and the safety of driving can still be guaranteed.
The ideal L9 equipped with the Hesai AT128 is to integrate hundreds of lasers on a small number of chips, which not only meets the high performance and high resolution brought by the high number of wires, but also brings an additional advantage, that is, to turn lidar into a semiconductor product to ensure that lidar has a sufficiently elegant and compact size.
02
8MP camera into the mainstream?
Keyword two: 8 megapixel camera.
The ideal L9 uses a high-performance camera as the primary source of perception, and the vehicle includes six 8-megapixel cameras and five 2-megapixel cameras.
More than half of the cameras on board the vehicle reach 8 megapixels.
What's the point of this 8-megapixel camera?
From the product level, the application scenarios of the 8-megapixel vehicle camera include: weekly view (front view, side view, rear view), surround view, electronic rear view mirror camera and cabin camera.
At present, the mainstream application scenario is the front view of the vehicle, because the front view needs to undertake more complex recognition tasks, such as identifying traffic lights, road signs, etc., and more needs the support of high-pixel cameras.
The requirements for other angles are relatively low, and the task of identifying the target is relatively simple, and the main application at this stage is the 200-500 pixel camera.
For the battle of the inner volume of the 8-megapixel camera, as early as 2021, it has become one of the indispensable hardware for various car companies to enter the automatic driving.
The first to eat crab is the ideal car.
In 2021, the launch of the new Ideal ONE makes Ideal the first car company in the world to mass-produce an on-board 8-megapixel camera.
This time, the ideal L9 in the flagship model uses an 8-megapixel camera, which is not surprising.
What's more, this configuration is also the trend of the times, including WEILAI ET7, WEILAI ET5, Xiaopeng G9, Extreme Kr 001, Gaohe HiPhi Z, Lotus Eletre and other high-end models are equipped with 8 million pixel cameras.
There is also a big background here that with the improvement of the level of automatic driving, the requirements for cameras in autonomous driving systems are also getting higher and higher:
The original camera was only used for reversing assist and driving record, and the resolution of hundreds of thousands of pixels can meet basic needs.
However, with the expansion of autonomous driving functions and application scenarios, the focus of camera applications has shifted to assist systems for perceived surrounding environment during driving.
The 8-megapixel camera means that the higher the resolution, the farther and smaller targets can be detected and recognized, and it also means more secure.
For example, the previous camera:
The height of the 1-2 megapixel camera can only achieve a horizontal field of view of 50° and an effective detection distance of 150 meters.
The 8-megapixel camera is capable of rising to a horizontal field of view of around 120° and an effective detection range of 200-250 meters.
For the ideal L9, the combination of six x 8MP cameras + five x 2MP cameras provides the following benefits:
360° all-round perception of the body's surroundings and long distances can be realized;
The combination of two 8-megapixel cameras directly in front enables the identification of objects such as vehicles, pedestrians, and cone barrels with a wide-angle range of 120° and beyond.
03
The higher the computing power, the more confident it is?
Keyword three: 508 Tops hashrate.
After talking about lidar and cameras, under such a strong hardware foundation, it naturally raises the requirements for computing power.
Even Weilai CEO Li Bin also said that horsepower plus computing power is a new standard for defining high-end intelligent electric vehicles.
NVIDIA CEO Huang Jenxun has also expressed similar expressions, Tops is the new horsepower, and the higher computing power of the car means more intelligent.
The ideal L9 intelligent driving computing power platform is equipped with two NVIDIA Orin X chips, and the final force has reached 508Tops.
In contrast, the current highest computing power and mass production delivery is still WEIO ET7, 4 NVIDIA Orin chips finally have a power of up to 1016Tops.
What is the use of such a high computing power?
Here we need to clarify a point of view: the level of computing power is only a hardware parameter, just a theoretical upper limit, the real utilization depends on the cooperation with the software algorithm, software and hardware need two-wheel drive.
This is why in this issue of spoilers, Ideal also emphasizes the "full-stack self-development capability" - perception, decision-making, planning and control software, all self-developed. (For full-stack self-developed ideal empowerment geometry, if you are interested, you can read this article: "Self-developed Miracle" of the ideal car)
In addition to meeting the needs of software such as cameras, perception algorithms, control execution capabilities and other software aspects, the main purpose is to embed hardware in advance for future OTA upgrades.
"Hardware embedding, software upgrade" is also the mainstream strategy of car companies for the life cycle management of new models - through the embedded large computing chip to ensure the continuous software upgrade capability of the vehicle in the whole life cycle.
Overall, compared with the ideal One, the ideal flagship model, the ideal L9, realizes the navigation assistance driving function of all scenarios by turning the high-end sensor and computing platform of automatic driving into standard, with a fully developed software algorithm.
With the "spoilers" of the ideal L9 for four consecutive episodes: exterior interior design, new paint color, new wheel shape, part of the cockpit configuration, part of the auxiliary driving configuration, new interaction mode, power system configuration, chassis system configuration, electrical and electronic architecture, intelligent driving system, body architecture, etc., a full-size SUV full of technology is gradually and clearly displayed in front of our eyes.
According to the time when the Ideal L9 will be officially released on April 16, the Ideal L9 probably has two more episodes of content updates.
In the next two spoilers, what information will the Ideal L9 release to the outside world?