On November 17, XPeng released an article explaining the difference between highway NGP, City NGP, and XNGP.

With the launch of Xpeng G9, an “ultra-fast charging and fully intelligent SUV” and the completion of the 4th 1024 Science and Technology Day Conference, the XNGP intelligent assisted driving system has been officially released, and it has the potential to become the top configuration of intelligent assisted driving before reaching autonomous driving. However, recently we have often heard XPeng friends say: XNGP, highway NGP, and City NGP are all called NGP. What is the difference?
Today we will explain it.

NGP — Intelligent navigation assisted driving

Speaking of similarity, having “NGP” in their names is the biggest and only similarity between them. So, what exactly does NGP mean?

NGP=Navigation Guided Pilot, the English literal translation is navigation assisted driving, that is, when the user sets the destination on the Xpeng vehicle and initiates navigation, and then activates the function by pressing the lever down twice, the vehicle will automatically follow the navigation route to the destination land.

Of course, since we are still in the stage of “human-machine co-driving”, the driver should also keep his attention and hold the steering wheel during the function operation, and when necessary (For example, when the car screen or voice broadcast prompts to take over, the traffic environment is complicated, etc.), timely take over.

 

Highway NGP — Applicable to NGP on highways and urban expressways

After understanding the meaning of NGP, it is basically clear what highway NGP and City NGP are. As the name implies, highrway NGP refers to the NGP function that can be run on most expressways and some urban expressways covered by advanced driver assistance maps.

The current applicable models are Xpeng P7 E version/P version, Xpeng P5 E version/P version, Xpeng G9 Pro version/Max version with corresponding software services
(The XNGP system is reflected in the highway NGP capability on highways /expressways covered by advanced driving assistance maps)

 

City NGP — NGP suitable for urban main and branch roads

City NGP refers to the NGP function that can be run on some roads in the city covered by the advanced driving assistance map. It is also another star function based on the mass production of Xpeng’s first-generation intelligent driving assistance system XPILOT.

The current applicable models are Xpeng P5 P version and Xpeng G9 Max version with corresponding software services (The XNGP system is reflected in the City NGP capability on the main and branch roads of the city covered by the advanced driving assistance map)

Although the City NGP and the high-speed NGP are “siblings of the same mother”, in addition to different applicable scenarios, they are also “very different” in many aspects!

1. Different hardware equipment

In urban working conditions, there are common problems such as mixed flow of people and vehicles, pedestrian crossing, short-distance cut-in, special-shaped roadblocks/traffic participants, etc., and the ability of laser radar to detect the distance accuracy and outline of objects (especially short-distance) far exceeds other perceptions device. Therefore, under the current circumstances, only models with lidar can realize the City NGP function.

2. Different technical architectures

Key points: highway NGP + urban scene ≠ City NGP!
Although they are all navigation-assisted driving, because the complexity of urban traffic is multiples, if the difficulty value of highway scene traffic is “primary school”, then the difficulty value of urban scene traffic is comparable to “university”.
Therefore, the City NGP adopts a brand-new technical architecture, including the comprehensive improvement of positioning, perception, prediction, regulation and control capabilities.

In the past year, in order to open up the urban scene, our self-driving teams from eight places in China and the United States used the combat method of “day and night, all year round”, and continued to polish the functions from the three aspects of “safety, usability, and ease of use”, and the software iteration exceeded 500 versions.

Compared with high-speed NGP, the code volume of City NGP is 6 times, the number of perception models is 4 times, and the code volume related to prediction/planning/control is increased to 88 times, which is by no means a simple scene adaptation.

3. Different user meanings

Did you know that in daily life, only 25% of users pass through expressways every day, but 100% pass through urban roads. The data shows that urban road mileage accounts for more than 70% of the total mileage traveled by users. It can be seen that an City NGP that can cover daily high-frequency traffic scenarios in the city is a “just need” for users, and this is also the core reason for our “overhills and mountains to fight tough battles”:
Only by solving the commuting problems in urban scenes can 90% of users travel and live simpler and easier.

In addition, limited by hardware and computing power, both highway and City NGP need to be enabled in areas covered by advanced driver assistance maps. Therefore, highway NGP can cover most of the highways and expressways across the country. City NGP is only open in the Guangzhou area for the time being. We are also actively promoting the opening of more cities, such as Shenzhen and Shanghai.

 

XNGP — Xpeng’s new generation of intelligent driving assistance system, which is an upgraded version of XPILOT

The XNGP intelligent driving system can realize continuous all-scenario intelligent assisted driving under appropriate conditions, and is currently applicable to the Max version of the Xpeng G9.

The letter X stands for XPENG, and we chose to keep the word “NGP”. From a technical point of view, it is because the ability of XNGP also requires the user to initiate navigation and then activate it, continuing the method and use logic of Navigation Guided Pilot , and highway NGP and City NGP have been used by so many friends, it sounds good and easy to remember!

However, XNGP and XPILOT are really different! Whether it is software or hardware, there are intergenerational differences in technology. Models equipped with XPILOT cannot be upgraded to XNGP through software OTA. At the hardware level, the current necessary conditions for XNGP are: lidar + 8-megapixel binocular camera, and a computing power of more than 500 TOPS.

What is the big trick of XNGP?

Available all over the country: Enable XNet deep visual neural network, heavy perception and do not rely on advanced driver assistance maps, can run with or without maps, so it can be used in all cities and all road sections in the country;
Can be used every day: 100% coverage of daily commuting, even in areas that are not covered by advanced driving assistance maps, you can complete actions such as turning around, changing lanes to overtake, bypassing obstacles, and avoiding pedestrians and non-motorized vehicles;
It can be used in the whole process: from the starting parking space to the terminal parking space, it can seamlessly connect various scenes such as highway, urban, underground parking lots, etc., the function will not exit, and users can enjoy “uninterrupted” smart driving experience from the start;
Comparable to an experience driver: In the highways and urban expressways covered by advanced driving assistance maps, it can achieve close to zero takeover performance;
In terms of overall safety, road competing ability and traffic efficiency in all traffic scenarios, it exceeds most drivers.

Caution! The XNGP system equipped with the G9 Max version has not yet enabled lidar, and the release plan for related capabilities is as follows:

In 2022, it will have highway NGP, ACC/LCC, memory parking, smart parking and other functions upon delivery;
In 2023, City NGP will be supported in cities with maps in Guangzhou, Shenzhen, and Shanghai, and traffic light recognition and the ability to pass, change lanes, overtake, and turn left and right at intersections will be gradually opened up in all cities without maps across the country;
In 2024, it will successfully evolve to the ultimate form, realizing the full-scenario intelligent navigation assisted driving ability from parking space to parking space.