Author: Zheng Senhong

If it’s already common for car manufacturers to offer free OTA updates and upgrades to users, is it rare to give away cockpit chips for free?

This time, the cockpit chips of over 25,000 Zeeker 001 car owners were paid for by Zeeker, costing over 300 million yuan.

On July 11, at the Zeeker Evolution Day, when Zeeker CEO An Conghui announced that the Zeeker 001 would officially be equipped with Qualcomm 8155 smart cockpit chips and that all users would receive a free upgrade, the barrage of comments was “Cong Cong YYDS”.

The car owners at the event and those online were all excited, pushing the atmosphere of the Zeeker Evolution Day to a climax.

Based on the current delivery and order numbers, Zeeker’s investment in the new generation of smart cockpits has exceeded 300 million yuan, covering hardware procurement, software development, logistics, and after-sales service to ensure that both old and new car owners can switch to the Qualcomm 8155 chip.

In terms of upgrading to the Qualcomm 8155 chip, NIO and Leading Ideal have previously given different solutions. Leading Ideal’s upgrade fee, including cameras and other hardware, is 4,999 yuan, while NIO’s starts at 9,600 yuan.

On this point, Li Bin said on the NIO APP, “More than 200,000 vehicles cannot be upgraded for free. More than 20 billion yuan cannot be supported.” This also prompted NIO’s old car owners to joke that Zeeker was not following the rules of martial arts.

In addition to the newly upgraded smart cockpit, Zeeker has also brought comprehensive upgrades to intelligent driving, chassis, and body. Offering free upgrades of both software and hardware, Zeeker can be said to have created an industry precedent.

ZEEKER OS 3.0, a “Big OTA” for the ages

Before ZEEKER OS 3.0, Zeeker had already updated its software versions four times via OTA:

• On November 5, 2021, ZEEKR OS 1.02 version involved 12 ECU units and 129 functional optimizations;

• On November 30, 2021, ZEEKR OS 1.0.4 version involved 13 ECU units and 187 functional optimizations;

• On February 12, 2022, ZEEKR OS 1.1.1 version involved 34 ECU units, added 25 functions, and more than 220 functional optimizations.- On March 23, 2022, ZEEKR OS 2.0 version involved 33 ECU units, added 5 functions, and optimized 59 functions.

• After 4 version updates, the software shortcomings of Jidu Auto’s first model Jidu 001 have been significantly improved. However, the sluggishness issue of the central control system has always been a temporary solution.

• Jidu Auto has finally come up with a perfect solution to this issue that has been criticized by car owners since the delivery of Jidu 001.

• Data speaks for itself. After upgrading to the 8155 cockpit chip, Jidu 001’s central control system performance has been further improved with the following enhancements:

• 2-second wake-up time

• Average 1.3-second turn-on time for in-vehicle applications

• 66.1% improvement in slide frame rate

• 3x faster cold boot speed of the central control system

• As the world’s first mass-produced 7nm process in-vehicle chip, Qualcomm 8155 is currently recognized by the industry as the most powerful cockpit SoC chip for mass-produced cars – sweeping the entire intelligent cockpit market and even winning almost all mid-to-high-end intelligent cars in China.

• Compared to the previously adjusted Qualcomm 820 that was only adjusted for stability, the biggest change for 8155 is the restructuring of the entire SoC structure and computing power, with notable improvements in fluency and ecological expansion.

• Prior to the official release of ZEEKR OS 3.0, we experienced the Beta internal test version:

• Fluency is not only reflected in the operation, but the voice interaction also brings visible and controllable functions to define the voice tone.

• Thanks to the computing power behind 8155, Jidu 001’s central control system interface has also been optimized to be more intuitive and straightforward. For example, a simple screen layout reserves the core position for high-frequency operations and simplifies the function level structure.

• In terms of ecological application experience, Jidu Auto has also improved the user experience based on their needs, especially in the car-phone interconnectivity. The user can easily connect to the car system for GPS navigation experience, and the destination can be automatically sent to the user’s phone.

In summary, after replacing the 8155 cockpit chip, the car infotainment system of Zeekr 001 runs faster, features smoother human-machine interaction, and provides a more immersive cabin experience.

Next is the intelligent driving part.

In the previous 2.0 version, Zeekr opened the ZAD advanced driving assistance functions, including Adaptive Cruise Control with Queue Assist (ACCQA) and Front Cross-Traffic Alert (FCTA), which are longitudinal control functions.

However, for the lateral control function like lane keeping, the version only provides Lane Departure Warning (LDW).

Therefore, the Zeekr 3.0 version has added the LCC lane centering assistant for lateral control, which is unique in the face of China’s complex road conditions and can demonstrate “smart” behavior, such as identifying and slowing down to account for pedestrians, displacing to avoid large vehicles, and slowing down and circumventing obstructions.

During the beta testing phase, Zeekr conducted simulated tests for pedestrian and sudden door-opening scenarios involving other vehicles. Based on the whole-car 7 8-megapixel long-range cameras and one 250-meter long perception millimeter-wave radar, Zeekr 001 can efficiently handle such sudden scenarios.

Finally, the chassis is updated, and the dTCS distributed traction control system is improved.

Compared with conventional TCS, which detects vehicle skids by wheel speed sensors with a slow response and long delay, the dTCS system, which is more suitable for new energy vehicles, moves the traction control system to the motor controller, shortening the torque transmission path and improving response speeds.

Therefore, after upgrading the dTCS system, Zeekr 001’s performance on extremely slippery surfaces is significantly improved, and the same solution can be observed on the BYD Dolphin.

Free is undoubtedly the biggest advertisement.

With a free soft and hardware upgrade of 300 million yuan, Zeekr gained not only the favor of new and old car owners, but also an OTA upgrade that exceeded user expectations and made up for the software system’s shortfalls, from driving experience to cabin experience.

When will the next stage of Zeekr ZAD arrive?Compared to Tesla, XPeng, NIO, and other new players, Zeekryi’s progress in assisted driving is somewhat slow, but this is due to Zeekryi’s choice of a “steady route” instead of a radical one.

To ensure the integrity of the active safety feature validation cycle, Zeekryi has adopted a phased OTA upgrade approach for the ZAD intelligent driving assistance function.

• Phase 1: Achieve lane centering and active cruise control, and continuously improve cruise function;
• Phase 2: Realize autonomous navigation in high-speed and urban expressway scenarios and some parking assistance functions;
• Phase 3: Continue to expand navigation and parking assistance scenarios to bring a better driving experience.

Now, the first phase has achieved its goal through two OTA updates, and the remaining second and third phases undoubtedly correspond to the city, high-speed, and parking navigation assisted driving scenes.

At the Zeekryi brand conference in April last year, An Conghui stated that a city autonomous driving solution will be launched in 2022, allowing autonomous driving to extend beyond the single high-speed road conditions, including features such as NZP Urban Navigation and TLB Red Light Brake Assistance.

Zeekryi’s intelligent technology model director Ye Jinyu said to Automotive News that regarding Zeekryi ZAD’s feature opening progress, ZAD Enhancement Package and ZAD Complete Package will undoubtedly be launched this year.

Ye Jinyu did not disclose the specific launch time, but based on Zeekryi’s progress of three small-version updates in less than three months and two major-version updates in less than four months, the next major-version update may be launched as early as October this year.

As for why there are delays?

Ye Jinyu said that the slow progress of Zeekryi ZAD, and why common L2 assisted driving functions are not available immediately, is mainly due to coordination with suppliers.

Zeekryi ZAD currently uses the “Supervision Pure Vision City Assistance Driving” system, built based on the Mobileye EyeQ 5H chip.

Among them, Mobileye is responsible for perception data, and Zeekryi is responsible for rule and control algorithms. The key reason for the delay lies in the localization of perception data.# Translation

Ye Jinyu said that the success of the perception system is crucial to whether ZAD from Xpeng P7 can be adapted to the road scene in China, especially with the complex road characteristics involving pedestrians, electric bicycles, and large vehicles. Therefore, the teams need to make step-by-step software adjustments to create perception data that suits the domestic road scenario. After spending three to four months on software and road testing of ACC, and related adjustments on LCC, Jikua has started working on the adjustment of city NGP.

According to reports, Mobileye’s core researchers have also joined Jikua’s Hangzhou Bay headquarters, and other ZAD functions are expected to be launched shortly.

So, is the eagerly anticipated ZAD worth the wait? In the first phase of ZAD, the Adaptive Cruise Control with Queue Assist (ACCQA) and the Front Cross-Traffic Alert (FCTA) perform as well as mainstream assistive driving systems in scenarios such as congested roads, early identification, and active avoidance. For example, in ACCQA mode, the system can maintain continuous and automatic start-stop while driving in congested roads, even when the vehicle is moving at an extremely slow speed of 1km/h.

In addition, the newly added Lane Change Control (LCC) function can flexibly handle emergency situations such as sudden changes in road conditions, and can briefly exit the LCC mode even when the headlights do not support lane changes, until the user completes the lane change and the system automatically resumes LCC mode.

ZAD’s first-phase functions may be just common assistive driving functions, but its actual performance is comparable to mainstream systems in the market and can lay the foundation for higher-level assistive functions.

Jikua’s next model: proprietary algorithms, Nvidia chips, and lidars

In addition to betting on Mobileye’s EyeQ 5 chips, Jikua had been regarded as having the computational ability, efficiency, and software/hardware combination to compete with the mainstream systems, but this required costly development time. Based on this passive situation, Jikua will take a self-developed route in the future.

Starting from the next model, Jikua will plan to adopt a combination of Nvidia chips, lidars, and self-developed algorithms, in addition to the supplier model used by Mobileye. Nvidia provides a more open platform and a less restrictive form of cooperation, while lidars can provide more reliable perception signals in the face of China’s more complex road scenario.In terms of R&D, the core autonomous driving research and development team of Jidu Auto has nearly 500 members, with more than 100 self-developed personnel responsible for the supplier mode, and more than 300 research and development personnel responsible for the self-developed mode. They have now the ability to integrate parking, NZP, NOP, and urban autonomous driving scenes.

Transitioning from suppliers to self-research, Jidu Auto is not the first. Faced with the risk of being subject to the supplier mode, self-research has become the preferred route for more and more leading players.

For automakers, self-research means having more initiative and growth, first of all, they can completely control the entire development process, and secondly, they can proactively update and iterate.

With the support of Geely, Jidu Auto has stood on a high starting point since its inception.

In addition to having the ability to self-develop in future software and hardware full-stack, such as the three-electric system and SEA architecture, Geely also empowers Jidu Auto in the area of smart internet of vehicles, autonomous driving, and other intelligent ecosystems.

A free software and hardware upgrade not only gains the favor of new and old car owners, but also has the potential to help Jidu Auto reverse the current situation and re-enter the top position in the smart car race.

This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.