Currently, the development of autonomous driving is in full swing, but technologies that are still in the theoretical and conceptual stages account for a high proportion in the industry. Many technologies still seem to have some way to go before they are truly implemented. As a company that has always adhered to the "user-centered" concept, Great Wall Motors has always been aiming to implement technology to truly empower users' travel life, and this is also the case in the development of autonomous driving technology.
Of course, real power still needs real technology to prove. Today, we will take a look at the "big news" of Great Wall Motors Coffee Intelligence in autonomous driving technology.
The first CSS model to promote the autonomous driving process and the first large-scale application of the Transformer architecture in China
Through years of continuous R&D and innovation, Great Wall Motors has now made a major leap in the development of autonomous driving technology. Great Wall Motors He has also achieved a large number of breakthroughs in algorithms, data processing and other technologies. For example, it is the first to create a dual-cognitive intelligent model (collaborative scene safety cognitive intelligent model) that can cope with traffic conditions with Chinese characteristics and uses the cognitive algorithm CSS model as the core. It uses an expert library method to define autonomous driving scenarios and combines scene safety Factors, applying a data-driven approach to solve the regulation and control problems of autonomous driving, generally speaking, can avoid safety risks in advance and make the autonomous driving system safer. At present, Great Wall Motors' expert library has accumulated tens of thousands of scenarios, covering all road conditions such as "highway main roads - urban open roads - urban expressways - up and down ramps - intersections" during the autonomous driving process. Under different traffic scenarios, It can perform reliable autonomous driving strategic actions, such as giving way to pedestrians and automatically changing lanes.
In order to support a large amount of perceptual inference calculations, Great Wall Motors has launched the ICU 3.0, the mass-produced autonomous driving computing platform with the highest computing power in the world. The single-board computing power of the platform reaches 360TOPS, and can be continuously upgraded to 1440TOPS. It can be called the "most powerful brain" of autonomous driving. It can effectively support the on-board computing of large AI vision models and the screening, cleaning, desensitization and reflow of vehicle-side sensing data. To put it simply, it can be based on the external environment perceived by the vehicle, etc. A series of data are efficiently processed and can support 6-channel Gigabit Ethernet at the same time. The inter-board data transmission capacity reaches 6Gbps. It can access up to 14 channels of 8-megapixel high-definition cameras, 8 channels of high-resolution millimeter wave radar, and up to 6 channels of high-resolution millimeter-wave radar. Road solid-state lidar can fully meet the needs of autonomous driving in multi-scenario road conditions at current L3 and subsequent L4 and above.
What is more worth mentioning is that Great Wall Motors has now made it more advanced, with stronger learning capabilities and larger scale
Data, stronger anti-interference capabilities, and higher recognition accuracy The visual recognition technology Transformer neural network architecture is applied to products to improve the safety and stability of autonomous driving. Currently, there are only two car companies in the world that have put the Transformer architecture into large-scale use. Great Wall Motors is one of them, and it is the first in China. The high precision of the Transformer architecture relies on the input and processing of large-scale autonomous driving data. There will definitely be a massive output of autonomous driving data, which will accelerate the leap of artificial intelligence from perceptual intelligence to cognitive intelligence, and Great Wall Motors is exactly that A pioneer on this road.
Six major safety redundancies build the "strongest security team" to implement user safety with technology
The most important prerequisite for the true realization of autonomous driving is safety. There is no safe autonomous driving It can be said to be empty talk. Great Wall Motors has also made breakthroughs in ensuring the safety of autonomous driving, including sensing redundancy, controller redundancy, architecture redundancy, power supply redundancy, braking redundancy, and steering redundancy. The six major safety redundant systems of Coffee Intelligence 2.0 achieve maximum safety guarantee from perception, decision-making to execution, implement user travel safety into practice, and truly promote the development of autonomous driving.
Among them, perception redundancy adopts a multi-source heterogeneous sensor solution, including the cooperation of multiple types of sensors such as visual cameras, millimeter wave radars, and lidar sensors, to achieve full vehicle coverage without blind spots and the strongest front coverage. , which can more effectively ensure the safety of autonomous driving.
The controller redundancy at the control end adopts a dual-controller architecture, and the backup controller can complete switching in as little as 0.02 seconds. Architectural redundancy includes a full communication architecture with 3 independent physical communication links, fully ensuring command transmission and reception. Power redundancy adopts dual power supply configuration, and the dual isolation solution can block all risks within 500us. The braking and steering redundancy part of the execution end also has obvious advantages. The braking redundancy adopts ESP IBooster dual brake redundancy, reaching a maximum full capacity coverage of -10m/s2, and 0-120km/h full speed domain control; while the steering redundancy It is the world's first fully redundant steering, with full-scenario failure mode response, and can achieve straight driving, turning, and U-turns with a minimum of 50 assists. The six major safety redundancies of Coffee Smart 2.0 can be called the "strongest security team" for autonomous driving. It can provide all-round security protection for autonomous driving at the software and hardware levels, and achieve 360° safety protection for the entire vehicle, which not only ensures ultimate safety but also further improves safety. Accelerate the implementation of autonomous driving.
Great Wall Motors’ Coffee Smart Six Safety Redundancies
Many of Great Wall Motors’ autonomous driving technologies have been implemented. For example, the autonomous driving technology using the Transformer architecture has been installed on Wei brand Mocha and Tank 300. Multiple safety redundant systems have also been applied to mass-produced models such as the city version and the first model of the new salon brand. In the future, more autonomous driving technologies will continue to empower more models of Great Wall Motors, which demonstrates that Great Wall Motors uses technology to change users. Attitude and determination in traveling life. In addition to powerful technologies on the algorithm and perception decision-making side, Great Wall Motors has also made major breakthroughs on the autonomous driving execution side, developing a smart wire-controlled chassis that supports L4 autonomous driving. Through all-round technological innovation, Great Wall Motors has rapidly promoted the development of autonomous driving and is also accelerating its transformation into a global smart technology company.