Intelligent driving "Street Competition": Who can beat Tesla, Great Wall, Huawei, or Xpeng better?

As the penetration rate of electric vehicles continues to rise, intelligence led by big models and big data has become the key to improving the competitiveness of automobile companies. Intelligent driving will lead a new round of changes in the automobile industry. This industry consensus was more fully reflected at this year's Chengdu Auto Show. In order to stand out among the new products, many car companies have focused on intelligence and used concepts such as "autonomous driving" and "intelligent operation" as promotional selling points. I hope In order to solve the current dilemma of homogeneous competition.

Among them, Wei Pai, a subsidiary of Great Wall Motors, demonstrated its AI self-driving capabilities at the Chengdu Auto Show and announced that with the blessing of the large self-driving model DriveGPT Xuehu Hairuo, Wei Pai is accelerating into the era of self-driving 3.0 and will soon launch a car equipped with 2 lasers Radar, version with urban NOH function. This means that Great Wall Motors is about to make efforts in the field of urban intelligent driving assistance in order to gain dominance on the intelligent track.

The first to adopt the "heavy perception" technology route, surpassing new forces on par with Tesla

In fact, since the end of last year, more and more models equipped with urban-level intelligent driving systems have been unveiled. At this stage, four major companies have announced that they have high-level urban assisted driving capabilities, namely: Tesla, Great Wall, Xiaopeng, Huawei. The intelligent driving solutions of these companies are very representative and basically cover several paths in the current intelligent driving industry.

One is Tesla, which takes a purely visual route, mainly using cameras to simulate human eyes to identify targets. Although this purely visual perception solution that abandons radar and high-precision maps can significantly save vehicle manufacturing costs and lower the threshold for smart driving to enter the consumer market, it still has problems that are difficult to avoid. For example, the lack of relevant radar will make it incompatible with more scenarios. , unable to adapt to complex road conditions: relying solely on the camera may cause scene recognition errors, and the camera may be "blinded" in dark light environments. Tesla has also caused many traffic accidents, leading to widespread controversy in the industry about this route.

Different from Tesla's purely visual technology route, Chinese car companies such as Great Wall, Huawei, and Xpeng have all chosen fused sensing solutions that are more suitable for China's complex road conditions. However, major car companies have also taken different paths on the path of integrated sensing. For example, in the early days, Huawei's ADS and Xpeng's NGP relied more on high-precision maps.

But there are a series of problems with over-reliance on high-precision maps. The first is that the coverage is limited, making it difficult to achieve the layout of all domestic cities and wider coverage; the second is that the freshness of the map cannot be guaranteed, and due to the complex drawing process, high-precision maps cannot synchronize temporary road changes in a timely manner, such as Temporary road construction, diversion, etc. At present, both Huawei and Xiaopeng have recognized these problems, and the new high-end intelligent driving systems have shifted to a technology route that emphasizes perception.

In this regard, Great Wall Motors has gone further.

Different from Tesla's purely visual route and new forces' over-reliance on high-precision maps, Great Wall Motors was the first to propose a "heavy perception" technology route, through a four-in-one highly efficient and collaborative super-sensing module (lidar + millimeter wave). Radar + ultrasonic radar + multi-camera) to obtain scene information to achieve dual protection of the visual system and perception system. And with a variety of sensing methods, it can accurately identify dark light environments or small objects. It can be said that Great Wall Motor's urban NOH not only combines the advantages of the above two routes, but also makes up for their shortcomings, and can adapt to more urban road scenes.

Great Wall launches a "combination punch" of software and hardware to create the optimal solution for urban assisted driving

With the support of high-performance hardware configuration, powerful computing power and autonomous driving data intelligent system, Great Wall Motors City NOH has created the optimal solution for urban assisted driving.

In terms of hardware configuration, the Wei brand Mocha DHT-PHEV lidar version currently equipped with the urban NOH system and undergoing generalized road testing is equipped with 2 125-line lidars, 5 millimeter wave radars, 12 ultrasonic radars, 4 Megapixel surround-view camera, 4 megapixel side-view cameras, 4 8-megapixel perception cameras, a total of 31 high-performance perception components. Both the quality and quantity of the perception modules are clearly ahead of the current mainstream assisted driving models.

In terms of computing power, Great Wall Motors has self-developed the full-stack IDC3.0 intelligent driving computing platform. The physical computing power of its domain controller main chip single board has reached an astonishing 360TOPS. In the future, it can be continuously upgraded to 1440TOPS through inter-board cascading. . You must know that the FSD chip developed by Tesla Model 3 has a total computing power of only 144TOPS.

Not only that, the algorithm is the trump card for urban NOH to stay ahead of its peers. Based on massive data, Great Wall Motors has built China's first autonomous driving data intelligence system, MANA, which focuses on the five major capabilities of perceptual intelligence, cognitive intelligence, annotation, simulation, and calculation, giving urban NOH the ability to continue to evolve, so that vehicles can behave like humans. Perception, thinking and decision-making. As of August 2023, the MANA system has completed more than 840,000 hours of simulation learning, and its virtual driving experience is equivalent to 110,000 years of human drivers. This not only provides strong data support for the implementation of urban NOH applications, but also allows urban NOH intelligent assistance Driving systems continue to grow, becoming faster, more stable, and safer.

At present, the first half of electrification has not yet ended, but the whistle of intelligence has already sounded in the second half. In this game, whoever can control the throat of intelligence will win the opportunity for development. To this end, based on the company's strategy and business development, Great Wall Motors officially announced the establishment of an intelligent cutting-edge organization - TCAL (Technology Center Al Lab, referred to as Al Lab), which is responsible for building Great Wall Motors' full-link AI technology system and building industry-leading AI capabilities. . It is believed that in the future, with the technology empowerment of Al Lab, Great Wall Motor's "heavy perception" capabilities, computing power and autonomous driving data intelligent system will continue to evolve rapidly, firmly establishing itself as the first echelon of intelligent driving tracks.