Global pilot, full push.
This year, the competition of intelligent driving has entered a "new stage". Taking NIO as an example, it has officially announced the global pilot assistance NOP+ urban function, and from April 30, it will also be fully pushed to users of Weilai Banyan Banyan Intelligent System.
To do a good job in intelligent driving, there is a road that cannot be bypassed, you need to continuously accumulate data in live scenarios, and constantly optimize and verify, which requires both manpower and time. "Without him, only the hand is familiar", the truth is the same. NIO's ability to push the full amount this time also shows that they have made pre-preparations and spent a lot of time to improve the intelligent driving experience. It is worth mentioning that at the same time as the release of the global pilot assistance NOP+, NIO also launched the "Intelligent Driving Points" system. Their goal is to account for 80% of the time spent on intelligent driving by 2025, making it 10 times safer than human driving, and truly promoting the ultimate vision of intelligent driving of "liberating energy and reducing accidents".
What kind of hardware architecture does NIO use?
On the NIO NT2.0 platform, the intelligent driving hardware has been raised to a very high level. The Aquila Aquila Super Sensing System contains 33 perception hardware, with LiDAR being the popular protagonist, and the ADAM NIO Central Computing Platform, which has a total computing power of 1016TOPS for intelligent driving, because it uses 4 Orin chips. Some people will say that the "volume" hardware is only superficial, and there is no volume before the "volume" hardware. You have to think like this, hardware is the foundation, it is more like the "weapon" of the intelligent driving system, without the weapon at hand, no matter how strong your martial arts cheats are, it is difficult to show them. Therefore, if the hardware is raised, the vehicle can obtain more accurate and comprehensive perception information, and the processing system can quickly carry and calculate the required results.
Moreover, in terms of algorithms, NIO is also not far behind. NIO's intelligent driving system can cover all application scenarios, and its general generalization capability must be the foundation, and the unified algorithm architecture is the key. To this end, NIO has created the NADArch intelligent driving architecture, which includes advanced algorithm model applications such as Lane 2.0 perception network, NADCloudM, and NADHVH.
In the above algorithm model, Lane 2.0 can support real-time perception of traffic at intersections in urban scenarios, NADCloudM can improve perception capabilities with the help of a stronger large model in the cloud, and NADHVN can make the global navigation assistance NOP+ have more delicate interaction capabilities through a fully data-driven planning network.
Not only good hardware and algorithm models are required, but also a large amount of live data. Each NIO NT2.0 platform model is a member of the "smart fleet", which can form a mass production fleet of more than 200,000 vehicles. Each vehicle is equal to a swarm intelligence and personalized training chip with a computing power of 254TOPS, and then forms a huge training network, with a monthly mileage of more than 20 million kilometers of actual vehicle verification, and a single route is automatically approved for 5 verifications, thereby accelerating the iterative upgrade of the intelligent driving system.
What is the scale of users and the scope of verification?
From April 30, the global pilot auxiliary NOP+ urban function will be fully pushed to nearly 230,000 users. This set of global navigation assistance NOP+ already has the functional experience of multiple scenarios such as urban areas, high-speed, and battery swapping.
In the high-speed road scenario, it can complete functions such as autonomous speed regulation, independent selection of the optimal lane, overtaking slow traffic, construction detour, active offset of large vehicles, and entering and exiting ramps according to the navigation route. In urban scenarios, it can complete functions such as autonomous speed regulation, independent selection of optimal lanes, overtaking slow vehicles, traffic light recognition and start and stop, and autonomously completing left turns, right turns, U-turns, and bypassing temporary parking vehicles, two-wheeled vehicles, and construction scenarios at intersections following the navigation route. On the day of the press conference, NIO also released a set of data: they have fully verified the road of the all-domain navigation assistance NOP+, with a verification mileage of 1207977 kilometers, covering 726 cities. Among them, 360,000 kilometers of high-speed highways and 847,000 kilometers of urban areas have been verified.
What is the "intelligent driving point" of people-vehicle co-driving?
Another piece of information worth noting is that NIO has launched the "Intelligent Driving Points" system, which is a new attempt based on "people-vehicle co-driving". "Intelligent driving points" is said to the driver. They combined the in-cabin face gaze detection and the extravehicular perception system, and modeled and evaluated the human-vehicle co-driving behavior from three aspects: focusing on safe driving, reasonably completing the system takeover response, and accumulating good driving experience, and took the past 30 intelligent driving days as a cycle, and gave a score, which can better guide users to drive safely and avoid accidents.
Intelligent driving is not completely unmanned. NIO is also aware of this, so their vision is to "free up energy and reduce accidents" to assist users in driving, not to replace it. "Intelligent driving points" are more like a communication carrier for people and vehicles to drive together. Let users better understand intelligent driving, and can also better use intelligent driving, which can not only liberate driving energy, but also reduce traffic accidents. Relaxed, safe, both have to have, both have both.
Intelligent driving, returning to a rational state, will develop better. New automotive technologies will also make our lives better and travel safer.
The author of this article is Kick Car Gang Xibei