How to evaluate Leapmotor’s NAC technology? How did NAC technology appear?

A technology launched by Leapmotor allows drivers to experience the fun of "Subject 2" driving. This driving technology called NAC can realize the system control of the accelerator and brake, and the driver controls the direction.

Netizens have mixed opinions about this. Some people think that this function is friendly to novices and the price is not expensive. Although the NGP applicable to the city's main and branch roads is new and exciting. You can park your car at the entrance of the parking lot with one click, which is easy and convenient. But if you want to experience these now, the price is high and the applicable range is small. Many Or futures.

Some netizens feel that this technology has little value, but it is packaged very bluffingly by car companies.

Regarding the "semi-open driving assistance system gameplay", in fact, not only Leapmotor has tried it, but controversy has always existed. Today I will introduce to you how these technologies emerged and their characteristics.

If you can't free your hands and feet at the same time

Then liberate one of them first

The essence of automatic assisted driving is to free your hands and feet, and the car controls the direction of the accelerator and brake.

We've seen autonomous parking as early as 1992, when Volkswagen unveiled the Futura concept car. According to the planning at the time, its implementation strategy was that the driver could get out of the car and watch the entire process of automatic parking. A computer the size of a personal computer was installed in the trunk to control the entire automatic parking system.

However, fully automatic parking has very high requirements for sensing hardware, central control systems, and execution systems. For example, the accuracy of the cooperation between millimeter-wave radar and cameras needs to be ensured, and the central control system's control of vehicle body speed and position accuracy is also particularly critical; The execution system must be able to ensure accurate steering wheel angle, throttle opening and other actions, which requires the electric power steering system to coordinate with the car's engine electronic control system.

If these problems cannot be solved in one go, is it possible to control some of them?

In Volkswagen's fuel-powered models, we have long experienced the service of handing over our "hands" to the car. For example, on the 2013 Tiguan, when we approach a parking space and the system detects a suitable parking space, the driving computer will prompt the driver to shift into R. This process may require several switches between reverse and forward to enter the parking space. Although the accelerator , brakes and clutch are still controlled by the driver, but at least his hands are not busy.

Compared with some brands of fully automatic parking, this technology does not look advanced, but in fact, even by 2023, only 7 of the 16 cars tested will receive a "G" rating, which is far lower than the performance of other systems. . This shows that some low-level parking assists are not actually to the point where they should be eliminated.

In addition to the method of giving "hands" to the car, there is also the option of giving "feet" to the car.

Leapao C10 launched a NAC technology (Navigation Assist Cruise) at the pre-sale conference. The system controls the accelerator and brake, and the driver only controls the direction.

The existing ACC algorithm can only realize simple actions such as stopping, starting, and following, and requires that there must be followable obstacles in front of the vehicle. It cannot react to turns, traffic lights, and braking in front of crosswalks at intersections; it can only respond to obstacles in front of its own vehicle, and cannot cooperate with the driver to intelligently accelerate and overtake.

Leapao NAC can be regarded as an advanced version of the traditional ACC function. It adds an acceleration and deceleration function to ACC. Through this function, Leapao C10 can combine navigation maps to identify zebra crossings, traffic lights, intersections, curves, etc. on urban roads. According to the road conditions, it can start, stop, accelerate and decelerate by itself to help users relieve driving stress.

Zero-run NAC solves the problem of traditional ACC's insufficient adaptability at intersections or curves, and the driver's frequent intervention causing the system to exit. Moreover, it has low hardware requirements, no threshold for opening a city, and lower cost. If you want some models under 150,000 yuan to quickly configure and get on the car, this model is worth considering.

If unmanned valet parking is expensive

Then implement customized parking first

For unmanned valet parking, you can give instructions directly through the mobile app, and the vehicle will automatically drive from the drop-off area to the parking lot and find a parking space by itself; when picking up the car, you only need to issue the command again from the app, and the vehicle will automatically return to the car. district.

Xpeng Motors' full-scenario automatic parking has launched the first shot in the intelligent transformation of automobiles. Car owners can realize this through key + voice control and 40cm parking distance + multi-parking space memory and selection, including side parking spaces, vertical parking spaces, diagonal parking spaces, unmarked parking spaces, etc., which can basically meet the parking needs in actual domestic car use scenarios.

Explained in the language of cars, cars need to have the ability to quickly build maps without pictures; when they sense that they are unable to pass, they will actively step back and give way; if they encounter pedestrians, they can also brake in time to protect them.

In order to reduce the difficulty, we can also try to dismantle some capabilities and realize at least some of the needs first.

For example, memory parking requires that before using it, you must first model the path and plan the memory route, set the starting point A and parking point B, and then you can park or summon according to the memory route. The route is fixed, and the parking point is memorized during the parking process. , the driver must be ready to take over at any time. For those who have a fixed parking space, this function is enough, and more advanced ones are just the icing on the cake.

Then there is customized parking. For example, on the Wenjie M5 smart driving version, automatic parking can be completed in any non-standard parking space without landmark lines. This has higher freedom than traditional automatic parking. Spend. I have to say that all the friends who park in the old community are happy. This feature is worth at least tens of thousands. After all, parking in corners is common. Usually, the car will be scratched if you are not careful. Now you don’t need to worry.

Custom parking spaces are a new feature of APA smart parking assistance. You can "mark" a temporary parking space for yourself by dragging the vehicle icon on the ground where parking is allowed but without markings.

If city coverage is difficult

Then first realize point-to-point automatic assisted driving

Autonomous driving assistance has always been a long-standing problem in urban streets. Now the company with the largest number of cities in the industry with fully open smart driving is Xpeng Motors. They have added 191 cities in the beginning of the year. Combined with the 52 cities that have been fully released before, the total The number of cities covered reaches 243. (Not all streets are fully covered and must be used according to XNGP prompts).

Judging from my experience with XNGP of Xpeng G6, most elevated road sections can be easily covered, but when driving on the road, only some main roads can be opened, and it cannot be opened when reaching some branch roads.

Just waiting is not a problem, so car companies have found ways to launch "AI driving" services.

It is characterized by autonomous driving in specific scenarios. Artificial intelligence will learn and store driving routes on designated road sections to achieve point-to-point assisted driving from home to work.

Many car companies are now launching this service. In addition to Xpeng's "AI driving", it also has the ideal commuter NOA function and Huawei's NCA function.

The ideal commuting NOA function allows users to set their own commuting routes, accumulate NPN (neural prior network) features through automated training during daily commuting, complete the training of simple routes within one week, and activate the NOA function on commuting sections. On more complex routes, training is expected to take 2-3 weeks to complete.

In the actual riding experience, the car can recognize most driving scenes and elements such as traffic lights, pedestrians, and vehicles in the scene, and make reasonable judgments in a timely manner. At the same time, the vehicle can drive stably when changing lanes, turning, and meeting other vehicles, and it can also perform exploratory operations similar to human drivers when encountering narrow road sections.

Summarize

Good technology must first have vitality and be able to promote the development of the industry. Secondly, it must be able to solve the urgent needs of users. Semi-open driving assistance systems seem to have castrated some capabilities, but that does not mean they should be killed with a stick. They are products that meet market needs and cost needs. They are not necessarily the most perfect or the most high-tech, but they do have their shining points.