eastwest connect

NIO’s autonomous driving future relies on full-stack self-developed NOP+, not Mobileye

NIO’s autonomous driving future relies on full-stack self-developed NOP+, not Mobileye


It is learned from Chinese auto media Bitauto, recently, NIO released two significant press conferences. One is the upgrade of the three major intelligent systems on December 2, and the other is the NOP+ media communication meeting on December 5.

The former elaborated on the iterative plan of NIO’s existing three car-machine systems, while the latter announced the push of the NOP+ assisted driving.

In the subsequent days, car owners who participated in the NIO user pilot group should gradually get the version update of their corresponding car system and the push of NOP+.

From smart cars to assisted driving, NIO, which was once leading and then behind, finally drove out of the corner and was catching up with full throttle.

NIO was once leading in assisted driving

In October 2020, NIO officially launched to ES6 and ES8 car owners the highway navigation assistance NOP function that can automatically enter and exit highway ramps and independently select lanes based on car navigation.

At that time, there were still more than three months before the launch of XPeng’s highway NGP.

The sales volume of Li Auto ONE had exceeded 20,000, and at the same time, it was involved in a car accident dispute where the assisted driving failed on highway.

Looking around the world, apart from Tesla, there was only NIO, a car company that could officially push highway navigation assistance functions to users.

Why NIO abandoned Mobileye’s solution

However, the partnership with the Israeli company Mobileye was blissfully short-lived. NIO originally planned to build L4 autonomous driving capabilities based on the NT1 platform and use Mobileye’s perception solution, but soon the industry realized that mass-produced ADAS and L4 functions are not a system at all, and it will never go beyond ADAS. L4.

NIO’s perception solution based on Mobileye EyeQ4, due to the limitation of computing power, the continuity of functions is not as good as internal expectations when facing scenarios such as entering and exiting highway ramps and automatic lane changes.

After more than two years and a total of 14 iterations, the capability of the NT1 platform is close to the bottleneck.

Moreover, compared with smart chip companies such as Nvidia and Horizon, Mobileye’s algorithm is more closed, and it is not open enough for secondary development of data.

All of the above prompted NIO to start thinking about how to achieve a point-to-point, cross-country, and sustainable iterative upgrade solution.

Obviously, the full-stack self-developed assisted driving and autonomous driving software and hardware system is the only answer.

NIO restarted with full-stack self-developed solution

On NIO Day in Chengdu in January 2021, NIO CEO Li Bin officially announced that NIO will restart the full-stack self-developed autonomous driving with L4 capabilities. NIO’s NAD, or NIO Autonomous Driving, will be officially known to users.

Along with it, NIO announced ADAM supercomputing platform based on 4 Nvidia Orin chips and the Aquila super-sensing system with lidar.

Brand new platform, brand new architecture, brand new thinking.
For NIO, the full-stack self-developed software and hardware system for assisted driving and autonomous driving under the NT2 platform is the basis for NIO to realize NAD. But NIO immediately encountered a dilemma – should NIO overthrow the thinking and definitions accumulated in the NT1 era, or based on previous achievements, first ensure that the new car has the NOP function, and then merge it?

NIO has chosen to overthrow and start over.

However, “restart” must be very painful.
Especially during the more than one year of NIO’s silence, the domestic high-end assisted driving situation has undergone earth-shaking changes. For example, Jihu, equipped with Huawei’s smart car solution, attracted a lot of attention with a full-time urban pilot assistant. Later, XPeng took the lead in announcing the implementation of the real urban NGP function in Guangzhou.

NIO, which was originally scheduled to launch NOP+ in the third quarter of this year, hastened to push the function in December.

NIO’s NT2 platform covers ADAS functions. But NIO’s thinking on NAD is far more than that. It is a full-scenario capability that can cover highways, cities, battery replacement, and parking, and can quickly expand for different scenarios and countries under a unified framework.

Ren Shaoqing, assistant vice president of autonomous driving research and development at NIO, said that “NOP+ is not an extension of NP, but the starting point of NAD.”

It can be understood that starting from NOP+, the underlying logic of NIO’s entire autonomous driving is completely different from the previous NIO Pilot.