Author: Lifelong Learner
Introduction
What will be the most popular feature on the autonomous driving track in 2022? The answer is none other than navigation-assisted driving. Navigation Guided Pilot (NGP) is an intelligent navigation-assisted driving system based on the user’s set navigation route, which enables the vehicle to travel from point A to point B. NOA (Navigate on Autopilot) led by Tesla abroad, and high-speed NGP by XPeng and NOP by NIO domestically, has shown intense competition in the current navigation-assisted driving market. This feature can effectively help drivers solve many daily problems, such as taking the wrong exit on a highway or overpass when they are not familiar with the traffic, queuing and merging at the on-ramp, and difficulties in changing lanes while driving at a high speed.
In addition to the NOA function that automatically enters and exits highways, autonomous valet parking (AVP) which solves the last kilometer is also the most concerned demand of users. Therefore, if domestic automakers want to have a say in the future intelligent automobile industry, they need to continue to break through the high-end intelligent driving and achieve end-to-end all-scenario intelligent driving. The all-stack solution of driving and parking integration will also become the best landing point for OEMs to complete the layout of the autonomous driving industry and achieve large-scale landing.
Driving and Parking Integrated Domain Controller is the Key Element for Domestic Navigation-Assisted Driving
What is the driving and parking integrated domain controller? Generally speaking, the development modes of the driving assist and parking assist functions usually adopt the separate development of driving and parking functions. However, this scheme also has the drawback of high cost. The driving and parking integrated domain controller can achieve both driving and parking assist functions through only one domain controller, greatly reducing the development cost.
The driving and parking integrated solution will be the mainstream solution for achieving high-end intelligent driving in the next few years. According to the data from the High AI Research Institute, the high-end intelligent ADAS market is currently in a high-speed growth period, and it is expected to usher in the scale production of the entire industry by 2022-2023. Among the mainstream intelligent driving suppliers, there are only a few who can occupy both the intelligent driving and intelligent parking markets at the same time. Although companies like Valeo and Bosch have a considerable market share in parking and driving, they have not achieved integrated design. And many Tier1 companies lack the accumulation of parking technology such as Continental Group, ZF, and Infineon, the market of driving and parking integration is still a blue ocean. KnowingTech is the most representative among the few mainstream intelligent driving suppliers who can occupy both the intelligent driving and intelligent parking markets at the same time.
Since 2016, Zhixing Technology has been implementing a full-stack strategy of software and hardware integration. Zhixing Technology has selected a progressive development route that is more in line with the production habits of automobile manufacturers. The company has taken intelligent forward-looking cameras and automatic driving central controllers as entry points, committed to achieving a high level of automatic driving in specific application scenarios first, and gradually expanding the application scenarios to achieve full-scenario intelligent driving. The current “driving-parking integrated” solution is based on the “driving-parking integrated” domain controller IDC mid as a high and low speed driving-parking integrated domain control solution that meets mainstream market needs. It provides customers with a leading L2+ intelligent driving experience and can cover driving needs in various scenarios such as high-speed elevated roads, urban roads, and parking.
Analysis of Key Technologies of the “Driving-Parking Integrated” Domain Controller
Currently, companies that have laid out in the “driving-parking integrated” field in the market include Zhixing Technology, Desay SV, Zongmu Technology, Zuishikj, Dongruirui, and Foryou Group. They have all launched their own “driving-parking integrated” solutions. Taking Zhixing Technology as an example, its current “driving-parking integrated” solution is based on the “driving-parking integrated” domain controller IDC Mid, which is a high and low speed driving-parking integrated domain control solution that meets mainstream market demands. IDC Mid provides customers with a leading L2+ intelligent driving experience.
Zhixing Technology’s IDC integrates both software and hardware. In terms of hardware, the flexible system architecture design supports compatibility with domestic chips. Functional safety is developed based on ISO 26262 ASIL B. It supports multi-sensor fusion and closed-loop development of cloud and big data, and also supports continuous iteration and upgrade of the extended high computing platform. It also provides rich interfaces, facilitating flexible configuration for customers. In terms of software, IDC Mid supports cross-platform operating systems and operating environments, providing an open development environment that enables continuous OTA upgrades. It will be mass-produced in Q3 2022 and used by several leading domestic automakers.
Based on this domain controller, Zhixing Technology has launched two versions: IDC Mid and IDC High. IDC Mid is equipped with a high and low speed driving-parking integrated domain control solution where driving and parking can share the same controller. It can achieve NOA (Navigate on Autopilot) navigation-assisted driving, L2++ advanced driving functions, HPA (Home Parking Assist) memory parking, and 360 panoramic functions through a 4R5V perception system. IDC High is an upgrade based on the IDC Mid platform, with a planned computing power of over 100 TOPS, and will deliver Urban NoA (Navigate on Autopilot) and AVP (Autonomous Valet Parking) applications in Q4 2023.
# Intelligent Parking with End-to-End Full-Scene Driving Drives the Development of the Industry
The intelligent parking system integrated with driving control (IDC Mid) developed by ZhiXing Technology, utilizing lightweight deep learning detection models for parking and lane lines, various types of lightweight deep learning detection models, multi-task deep learning detection models, and dynamic multi-camera calibration algorithms, achieves user-centered value experience through the advantages of perception algorithms. Moreover, based on perception algorithms, the system can also implement automatic and remote parking functions and efficiently accomplish automatic entry and exit of vehicles. The system supports different parking scenarios, including horizontal, vertical, inclined, spatial, line and brick parking spaces. In the same perception system, in case of sudden danger during parking, the system can quickly switch to automatic emergency braking mode to realize efficient and safe automatic parking.
In addition to satisfying various user needs, the integrated parking and driving control system has significant advantages over separate domain controllers on cost, overall performance, and functionality iteration and upgrade. In terms of cost, the integrated control system has reduced the number of controllers, and the driving and parking perception systems are integrated, lowering hardware costs from multiple aspects. As an example, the integrated parking and driving control solution offered by ZhiXing Technology can help original equipment manufacturers save 30% of costs. In terms of overall performance, the unified controller can effectively maximize the role of a single domain controller and achieve full coverage of perception from close to distant ranges, thereby making the intelligent driving more adaptable to various scenarios. Regarding the iteration and upgrade of intelligent driving functions, based on a unified hardware configuration and flexible software settings, the system can save development costs. According to related research, the integrated parking and driving solution will be the mainstream solution to realize high-level intelligent driving in the next few years. It is expected that the market will witness mass production of the solution between 2022 and 2023, and end-to-end full-scene intelligent driving will be achieved through OTA software upgrades around 2025, based on the integration of parking and driving control.
Leading Industry Development, End-to-End Full-Scene Intelligent Driving
Currently, mass-produced intelligent driving technologies mainly focus on parking at the beginning and end of trips and high-speed/highway driving during the middle stages. The entire full-scene intelligent function is still lacking, pulling together the pieces of the current intelligent driving and creating a new battlefield for automaker’s dominance in the next few years. Based on the domain control of parking and driving integrated system, L2++ driving and parking functions, hardware embedded technology and OTA remote software update technology, end-to-end full-scene intelligent driving utilizes a real-closed loop for verification and optimization to acquire, train, optimize and update algorithms for unknown-scenario data, constantly improving driving ability, and expanding the full-scene driving range for intelligent driving, ultimately achieving full-scene adaption.
Up to now, Know-First Technologies has landed multiple fixed-area L4 autonomous driving projects. Mass production and designated clients include Zeekr, Geely, Great Wall, Chery, Wuling, WM Motor, Geely SMART, etc. At the same time, Know-First Technologies has completed four rounds of financing with a valuation of 2.6 billion yuan, and is currently in a new round of financing. It is expected to complete stock reform in the fourth quarter of 2022 and submit an IPO application in the first half of 2023. The key to the success of a start-up autonomous driving company lies in whether it has been recognized by the market and investors. In the current environment, continuous capital investment is the guarantee for the industry’s productivity promotion. Continuous financing success also represents that Know-First Technologies is on the right track, and its solutions are gradually being recognized by the market.
Summary
Whether or not they have assisted driving functions has become a must-win battlefield for the next stage of intelligent driving car companies, and the ultimate goal of achieving fully automated driving is also the ultimate goal of various manufacturers. The integrated driving and parking solution may be an innovative way to break through the bottleneck. Obviously, Know-First Technologies has been deeply cultivating in this field for many years and has brought out its own mass production plan, which has also received orders from whole vehicle manufacturers, making it the leader in this field.
For most traditional car companies, self-researching navigation and assisted driving from scratch requires a lot of time and money. However, in the end, due to talent shortage or insufficient funds, they may not necessarily produce good products that are easy to use and have a high market recognition. Under the pressure of this risk, direct procurement from suppliers can not only quickly make up for this shortcoming, but also reduce the pressure and risks of research and development, making it a good choice. Know-First Technologies has a very rich product plan, and the most important thing is that their mass production solution has already been verified by the market. They already have high-speed L2 and city L2 in L2, as well as low-speed parking and L4 unmanned driving in low-speed closed areas, and is a fully-scenario intelligent driving technology company that traditional car companies may consider cooperating with. With the gradual increase in mass production orders from car companies, Know-First Technologies is already in a leading position among intelligent driving suppliers.
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.