As we all know, the development path of new energy vehicles is not like that of traditional fuel vehicles, which requires repeated technology accumulation and market verification to iterate to the next stage. Therefore, it only took a few years for the three-electric system of new energy vehicles to catch up with the bottleneck stage that the internal combustion engine took a hundred years to reach. At the same time, the homogeneity of products among different brands has become more and more serious. When it comes to new energy vehicle products, Today, with rapid iteration, if you want to make breakthroughs and acquire more customers, technologies such as intelligent driving assistance have naturally become a new battlefield for various car brands. Recently, Nezha Auto also disclosed the brand's phased progress in intelligent driving technology for the Nezha S model.
In the implementation stage of intelligent driving technology, the progress speeds of various brands are generally similar. The difference in the technical routes of domestic and foreign car-making brands is nothing more than a different degree of dependence on sensors, high-precision maps, graphics processing computing power, etc. Radicals will completely abandon various sensors and rely solely on visual learning and computing power optimization to improve driving assistance functions. Conservatives, on the other hand, rely more on the support of “hardware” such as millimeter-wave radar, lidar, and high-precision maps. At this stage, due to some restrictions in regulations and actual conditions, we cannot yet say which method is the absolute direction for the future. However, it is not difficult to see from the phased progress announced by Nezha Automobile that this brand is obviously the more radical among conservatives. of a wave.
Friends who have a little knowledge of driving assistance should have seen more or less reports on the integrated parking technology of brands such as Tesla FSD, Xpeng NGP, and Baidu ANP3.0. The main technical direction is driving assistance functions in high and low speed ranges, assisted parking, remote summoning, etc. So at what stage has the Nezha S, the latest product of the Nezha brand, reached in terms of integrated travel and parking technology?
Nezha S is currently on sale in a total of 8 configuration models, but there are two versions of intelligent driving assistance. One is NETA PILOT3.0, which is based on a dual TDA4 domain control platform + high-precision map + 7 environmental cameras + 4 surround-view cameras + 5 millimeter wave radar + 12 ultrasonic radar hardware combinations. There is also a NETA PILOT 4.0 that is based on the existing hardware of NETA PILOT3.0, replaces it with Huawei's MDC610 domain control platform and dual 8-megapixel autonomous driving front cameras, and adds two front-facing lidars.
NETA PILOT3.0 system is theoretically an L2+ driving assistance system with L3 driving assistance hardware. The TDA4 provided by Texas Instruments has a computing power of 16TOPS, which is higher than the 8TOPS computing power of Horizon J3, and the price is also cheaper. . Because the computing power reaches 16TOPS, TDA4 can also access more video sensors, millimeter wave, ultrasonic and other radar hardware to achieve a higher degree of automation.High driving assistance, ensuring driving safety and computing redundancy. In addition, its overall operating power is low, no complex heat dissipation structure is required, and it is even functionally compatible with high-pixel autonomous driving cameras. Some people may ask why TDA4 was chosen instead of Horizon’s J3. Officials said the main reason was due to misalignment of the product development cycle and adaptation issues in some subsequent feature development, so it was not ultimately installed on Nezha S.
NETA PILOT4.0 system is a configuration that theoretically has L4 level driving assistance hardware level. It uses the MDC610 chip from Huawei and has 200TOPS computing power. The hardware level can not only fully cover the needs of L4 level driving assistance, At the same time, the redundancy of computing power is also higher than that of TDA4. As for why Nezha S chose Huawei when many brands were choosing Nvidia's Orin-X, it is obvious that Huawei's system has already carried out in-depth cooperation with many domestic brands, and the local adaptation and procurement costs are obviously It has more advantages and is a better solution for a model like Nezha S with a price of around 200,000-300,000.
Although the hardware part comes from suppliers, the software level of Nezha S adopts a full-stack self-developed solution. The advantage of this is not only that it is not controlled by others, but also the core algorithm self-developed and data closed loop can be faster. Rapidly iterate drive products to enhance the experience of intelligent driving systems.
As for the integration of parking and transportation, Nezha has also released a point-to-point solution that covers all scenarios. The entire trip is from the starting point to the end point, covering urban scenes, high-speed scenes, tunnels, urban main roads or branch lines.
Nezha S, which has the NETA PILOT3.0/4.0 system, can turn on the NNP intelligent pilot assistance system on highway sections. The functions of NNP are similar to the existing high-speed driving assistance functions of many brands. But it has better speed control strategies and traffic efficiency, and high-precision maps will also optimize the accuracy of route and lane selection. In addition to wide-speed threshold adaptive cruise, follow-and-stop, lever lane changing, autonomous lane changing, vehicle avoidance, up and down ramps and other functions. After turning on NNP, the in-car screen will render lane lines, traffic participants, road conditions, water horses, signs, etc. in real time. NNP's functions can be turned on in various ways, and can be activated through levers and voice. Moreover, Nezha's system encourages users to use it on highway sections from the bottom up. Even when driving at high speed, it will actively prompt whether to use NNP.
We mentioned earlier that the driving assistance system of Nezha S is a more radical trend among conservatives, mainly because of its lane changing strategy. At present, the mainstream trend in the industry is to enable the vehicle to overtake the idle vehicle from the left after turning on the auxiliary overtime. The underlying logic of Nezha S's NNP will further optimize traffic efficiency on the basis of ensuring driving safety. It will judge more appropriate traffic based on road conditions.According to the appropriate route, a more efficient lane change strategy will be implemented. When conditions permit, the driver will independently choose to overtake on the left or right side. The experience will be more in line with actual human driving habits.
When merging into or out of a highway when using NNP, the system will flexibly change the speed of the vehicle and will not rigidly follow the current speed limit and drive at the same speed in the rightmost lane or ramp, affecting rear traffic. efficiency. Instead, on the premise of complying with traffic regulations, follow the traffic flow in front and behind to increase or decrease the speed appropriately.
At present, Nezha’s NNP function has begun to be gradually devolved to car owners. Friends of Nezha S can experience it while driving on highways. Of course, in addition to high-speed sections, the NCP city navigation auxiliary function in urban sections based on high-precision maps will be provided on Nezha S models with NETA PILOT4.0 hardware. High-precision maps, combined with lidar, 800W dual-camera and other hardware, can realize lane-level navigation in urban low-speed sections, identify traffic conditions at traffic light intersections and assist in passing, and other traffic participants can intelligently avoid obstacles when encountering them. This enables a series of complex road condition driving assistance in the low-speed range. For similar experiences, you can refer to the related content of Jifox Alpha S HI version and Baidu AVP3.0.
On some urban arterial roads and terminal roads without high-precision maps, Nezha S with NETA PILOT3.0/4.0 hardware can also enable smart driving on urban arterial roads. However, the official stated that it is still in the verification and testing stage, and the supporting high-precision maps are also being continuously optimized and collected. The trial version will not be gradually released until 2024 at the earliest.
In addition to driving, Nezha also released more information about parking this time. The memory parking function and smart summoning function will be decentralized next year. The memory parking function can perform memory parking within a distance of up to 1 kilometer, and supports learning and memorizing 100 routes. The parking process can be controlled through the mobile phone, and the car will actively avoid and bypass pedestrians, obstacles, other vehicles, etc. during the process. For example, this function is quite useful in some rainy and snowy days or when women are walking alone. Although we haven’t been able to experience it yet, combined with the WM PAVP and Xpeng VPA-L that we have experienced before, this feature is still good.
Smart summoning NMS can perform remote summoning on internal roads or parks. Strictly speaking, it is more like L4 autonomous driving. The summoning distance is enabled within a range of 500 meters. The system will plan the path in real time to achieve unmanned conditions. The vehicle can be operated to avoid traffic participants during the vehicle summoning process, with pedestrian priority as the core logic. This system is similar to Weimar's HAVP. Friends who are interested can check out our previous experience.
Nezha S performs well overall among medium and large new energy sedans, and is also remarkable in terms of price, experience, and performance.. With the successive decentralization of NNP, NCP, NMS and other functions, this model will achieve a qualitative breakthrough in intelligent driving and become a hexagonal warrior just around the corner. Of course, we will also conduct in-depth experiences with these functions in the future. Friends who are interested, please continue to pay attention.
Editor in charge: Xuan Yan