On this track, some giants have achieved mass production and commercialization, such as Gaode and NavInfo. Some have accumulated a lot of data and formed a solid moat. Giants have people, money and resources, and constantly seize market share, leaving few opportunities for startups. So how should startups break through when surrounded by giants?
Traditional graphic merchants and technology giants each regard mountains as king.
In 20 1 1-20 14, internet giants began to enter the electronic navigation map market. In April of 20 1 1, Baidu took the lead in launching the mobile version of Baidu map, and acquired Tushang Daotong in 20 13; Ali acquired Gaode in April 20 14; Only 1 month later, Tencent invested 1 1.7 billion in NavInfo, and acquired Lingke Hangrui in the same year.
20 14 When the graphic dealers were frequently acquired, with the emergence of autonomous driving and car networking technology, high-precision maps began to sprout, and technology companies and traditional graphic dealers entered high-precision maps one after another, constantly upgrading related products and services and expanding the "circle of friends".
Gaode is the first company in China to realize the commercialization of high-precision maps. In 20 19, gaode map announced that it would provide standardized high-precision maps to its partners at "cost price", and the average annual charge per vehicle would not exceed 100 yuan. Gaode established the UGC system of self-mining high-precision map and the automatic data production model. At the same time, it realizes the logistics distribution, vehicle-road coordination and high-precision map data update based on Ali Ecology.
Image source: Gaode map
NavInfo signed a license agreement on autopilot maps and related services with BMW China in early October19, which will provide map products and services for BMW's autopilot systems above L3. This is the first domestic order for mass production of high-precision maps of L3+ autopilot system, which also means that mass production of new high-precision maps of NavInfo has begun. NavInfo has independently developed a professional collection vehicle, and mastered the complete solution of data collection, mapping, crowdsourcing update and fast iterative autopilot map. At the same time, NavInfo also independently developed the FastMap production platform, and applied technologies such as image recognition and data mining to promote the collection and update of map data.
Baidu has signed commercial customization projects for high-precision maps with several automobile companies, including BAIC, Hyundai, Great Wall and BYD. At present, the automation degree of Baidu's high-precision maps has exceeded 90%, and the algorithm recognition rate has reached over 98%. The advantage of Baidu lies in its independent research and development of a set of production, production, release and update processes of high-precision map products. Based on Apollo open platform, Baidu has the self-research and self-control ability of high-precision map software and hardware equipment, and has complete intellectual property rights in collecting vehicles and software computing platforms.
According to Tencent's official statement, it has accumulated more than one million kilometers of road test data, more than 5 million kinds of labeled sample data, and has completed the production of high-precision map data of expressways and expressways nationwide. Compared with Arigaode, Tencent has its own complete autonomous driving scheme, which enables its high-precision maps to detect high-precision and errors in real time and quickly on Tencent's autonomous vehicles and simulation systems.
Image source: Siwei Tuxin
How do startups break the game?
Around 20 16, a number of high-precision map start-ups emerged at home and abroad.
Compared with giant companies, start-ups are facing the pressure of cost and resource shortage, and the speed of commercialization is also slow. If you want to make a breakthrough, you need to foster strengths and avoid weaknesses, and seize market opportunities that giants cannot cut into.
First of all, unlike traditional map merchants who have been deeply involved in navigation maps for many years, because each map collection vehicle is as high as several million and the cost is high, start-ups often adopt crowdsourcing mode, or cooperate with automobile manufacturers and travel service providers, similar to UGC mode.
Secondly, through differentiated product routes, start-ups can build their own technical capabilities. For example, in the way of collecting high-precision maps, in order to save costs, Wayz.ai, wide stool technology and so on. A pure vision scheme based on artificial intelligence technology is adopted. Compared with the lidar scheme of the giant company, the pure vision scheme takes the camera as the acquisition tool, which greatly reduces the cost.
Although the accuracy of the camera is not as good as that of lidar, in the long run, with the cooperation of crowdsourcing scheme and logistics team, the visual sensor can accumulate data through multiple shots, and the overall accuracy will probably be more improved than that of lidar.
On the other hand, the update frequency of pure vision scheme is higher than that of lidar scheme. At present, the map update frequency of giant companies is quarterly update of cities nationwide and monthly update of some cities. And startups that adopt pure visual solutions can achieve real-time updates in ten minutes, and even reach the update frequency of minutes in the future.
"The accuracy of map collection is on the one hand, and the update speed is the key factor that determines the outcome." A high-precision map entrepreneur told Geshi Automobile.
Although the giant has money and resources, the cooperation with the car factory is not smooth sailing. Through understanding, it is not difficult to find that there is more or less BAT behind the giants. For example, there is Alibaba behind Gaode, and both Baidu and Tencent have their own high-precision map teams, but this will worry car manufacturers.
When updating the map, the data on the car body is very important, which is helpful to upgrade the high-precision map. However, when cooperating with BAT, automobile manufacturers are not very willing to share data because they feel that "BAT is a competitor."
This also gives opportunities to startups. Car manufacturers often walk on two legs. On the one hand, they cooperate with high-precision map manufacturers, on the other hand, they invest in start-up companies and master high-precision map technology. For example, SAIC has invested in Zhong Haiting, Zhong Jing map and civil engineering at home and abroad. Start-ups such as maps and Deepmap.
In addition, both start-ups and original equipment manufacturers will face commercialization problems. Traditional cartographers make high-precision maps, and their mode is usually to cooperate with the main engine factory. However, some startups are autonomous driving solution companies that can produce and sell themselves.
Take NavInfo as an example, its main customers are automobile manufacturers. In other words, only when the car is sold can the company sell its products. In the past two years, the domestic automobile industry has been depressed, and the demand side has dropped seriously, which has also affected the income.
Although NavInfo won the world's first high-precision map order for self-driving vehicles above L3, it will not be available until the summer of 20021year, and the sales volume is unpredictable. The era of intelligent technology is different from the era of navigation maps, involving ecosystem security, higher threshold and stricter branding.
But in startups, such as Momenta, they are not only providers of map technology, but also users of maps. By applying the self-made high-precision map to the self-developed L3 and L4 autopilot solutions, the information accuracy of the high-precision map is redefined. At the same time, according to the needs of downstream customers, further improve the accuracy and update requirements of high-precision maps.
With the development of autonomous driving technology, start-up companies can also develop deep learning and vehicle-level computing chips, devices and ECU that support the rapid update calculation of a large number of map data. At the same time, it provides a solution for realizing multi-functional integration of software and hardware. Perhaps it is also an important development direction for startups.
Overview of the earth:
According to the prediction of Geshi Automobile Research Institute, the number of self-driving cars above L3 will increase significantly from 2020 to 2025, and by 2025, the number of self-driving cars will reach 3.67 million. With the development of autonomous driving, the demand for high-precision maps will also increase dramatically. It is estimated that the domestic high-precision maps will reach nearly 8 billion yuan in 2025, and there will be rapid growth in 2026, exceeding 654.38+0 billion yuan. At present, only 22 companies have passed the qualification of surveying and mapping, which is a huge market. Although the giants are deeply rooted, there is still room for development for start-ups. Start-ups should find new business growth points, strengthen cooperation, form their own advantages and accelerate the commercialization of their business.
This article comes from car home, the author of the car manufacturer, and does not represent car home's position.