1, more fine-grained. The map previously had administrative areas, transportation networks, points of interest data, and now the granularity is getting finer and finer. For example, indoors, Baidu Maps has invested in a specialized team to collect and maintain major shopping malls through backpack collectors for indoor navigation; then again, in this collection boot camp, Baidu Maps announced that it will conduct field panoramic collection of data on key natural ecosystems along China's coastline to record oceanographic landscapes; high-precision maps are the ultimate in fine-grained maps - -almost complete mapping of real-world traffic conditions, mainly for intelligent driving, so that the car has "eyes", Baidu Maps has set up a special team, equipped with a dedicated high-precision map collection vehicle.
2, more intelligent. AI to replace part of the human work is a big trend, the most likely to be replaced by repetitive work (including mental and physical), map collection in part of the work such as traffic sign recognition, POI identification, etc. can be done with machines. AI is Baidu's core strategy, AI technology is also Baidu's killer app, "Bring AI to Life". Bring AI to Life" Baidu is not only the AI landing to the product, but also focus on using AI to improve productivity, the most typical is the map data collection, with the help of image recognition, voice recognition, big data processing and other AI technologies, the formation of map data processing "three major knife-edge technology The most typical is the map data collection, with the help of image recognition, voice recognition, big data processing and other AI technologies, forming the "three major blade technologies" of map data processing, greatly improving the degree of automation of map data collection and processing, such as the application of voice interaction technology to achieve "one person, one car" in external operations, and the internal processing of intelligent identification of roads and POIs, at present, 80% of the work of the data collection link has been realized intelligently, the industry's leading.
3. More real-time. Vehicle-mounted offline navigation equipment data often need to be updated only once a year, many times the navigation route does not match the actual. The real-time collection of data in the mobile Internet era has become stronger, and real-time map data collection is a trend. Baidu map in real-time data can be said to have spared no effort: the introduction of UGC mode, encouraging users to report traffic accidents and other news; through big data to automatically analyze congestion; and cooperation with the traffic department to access real-time bus, high-speed rail, subway and other data ...... all kinds of initiatives, so that the map data become more real-time.
4, multi-tool synergy. Map data mainly with the collection vehicle, Baidu map has more than 250 collection vehicles, is the largest domestic map platform own collection fleet. However, users have more detailed requirements for the granularity of the map, such as the natural environment without access to highways, indoor shopping malls, but also need to directly see the panoramic map, AR navigation and so on. To meet these user needs, the collection equipment can not be limited to the collection vehicle, Baidu Maps this data collection training camp is a highlight is to show a lot of new collection equipment, such as the collection of backpacks, drones, hand-collected equipment, panoramic camera, etc., to realize the "land, sea and air" supporting some of the important places, such as popular scenic spots can be a panoramic panoramic collection.
A year's worth of data has been collected.
More than a year ago, Baidu Maps switched to its own intellectual property rights and stopped buying third-party services (excluding overseas markets), which means huge data collection and maintenance costs. Regular map collection and maintenance itself takes a huge amount of manpower and energy, and Baidu Maps is still sparing no effort to meet the increasingly complex collection needs of maps, which requires an even higher cost of investment. Up to now, Baidu Maps data collection, processing and R & D related personnel is close to 2,000 people, Baidu Maps Division General Manager Li Dongmin revealed in an interview with Tencent Technology, "Baidu Maps each year in the data collection accounted for about 40% of the total investment in Baidu Maps." , which believes that the basic data is the admission ticket to the map, "In the AI era, we still invest more than 50% of the energy in focusing on basic data."
Why is Baidu investing heavily in "unprofitable" business?
Initially, map navigation apps were charged, and third-party navigation devices had to pay to use map data. However, Baidu Maps' free measures in the mobile Internet era have directly changed the market pattern, and now the entire map industry has entered the free era. Therefore, at present, the Internet map business, are not scalable revenue capacity. For Baidu, the map is also a "cost-based" business, in other words, Baidu map is not like search, feeds, finance and other businesses to bring direct scale cash flow.
Why is Baidu still investing so much in its mapping business? I think the core reason is that it used to be Baidu's strategic defense business, because it is an important mobile portal; today, under Baidu's strategy of "consolidating the foundation of mobile, winning the AI era", the map has a higher strategic value: not just a defense business, but an infrastructure, and, in the words of Li Dongmin, is the water, electricity, and coal of the AI era.
First, Baidu All In AI strategy is based on data and algorithms, algorithms are Baidu Brain, NLP, Knowledge Graph, voice and other deep learning technologies, and data in addition to Baidu search data, the most important is the map data.
The map is an Internet mapping of the spatial level of the physical world, but also records the world's human, vehicular and logistics trajectories through LBS services, which is equivalent to caching the world. Baidu Maps has a total of more than 150 million POIs globally, provides location services more than 80 billion times a day, provides navigation services more than 300 million kilometers a day, and has more than 300 million monthly active users. Behind these figures is a large amount of spatial data, which lays the data foundation for the landing of the AI strategy.
Secondly, Baidu Bring AI toLife means that more and more scenes need map support.
Baidu is landing AI into more scenes, such as in-car navigation, self-driving cars, OTA (online travel), finance, Feeds, DuerOS, search and so on. All these businesses need map support, especially the self-driving car strategy, but also the support of high-precision maps, because high-precision maps are the eyes of the car. Baidu map is to support the above business, the value of this point with deep learning technology, natural language processing, knowledge mapping and other underlying technology - we never go to the luxury of technology to generate revenue directly, but the technology will drive the business can bring income.
This point is also the reason why Li Dongmin said that the map is Baidu AI strategy of water, electricity and coal, it can be said that the map is an infrastructure in the AI era.
Third, Baidu map is still an indispensable entrance for Baidu.
The frequency of map App use may not be very high compared to search, social networking, etc., but the use of navigation and other scenarios in a long time, especially occupying the car, travel and other gold scenes; there are also users who will use the map indirectly through many applications such as travel App, **** enjoy bike App, travel App and so on. Therefore, whether indirectly or directly, the map is an important mobile portal, Baidu in the "winning AI era" at the same time to "consolidate the mobile foundation", naturally focusing on the layout of the map as one of its important entrances! --In fact, Ali and Tencent also attach great importance to maps. Not long ago, Alibaba structural adjustment, Yu Yongfu no longer serve as the president of Golder, Ali Group Chief Risk Officer (CRO) Liu Zhenfei (Zhenfei) concurrently served as the president of Golder, Zhang Yong in the appointment of the mail, "We believe that Golder as a geographic location of the Internet service provider, will play an increasingly important and unique value in the economic ecosystem of Ali. "
It can be said that Baidu map for Baidu has a very strong strategic value, although it can not form a direct cash flow, but it is like AI technology, for Baidu's core business to play an indispensable role in supporting the accumulation of spatial data, for Baidu AI landing to each scene will unleash a huge potential for Baidu's a gold mine in the long term. Lu Qi to the positioning of Baidu map is "moat", its value is to ensure that the main channel business such as AI, Feeds sailing more robust, now it seems that Baidu map is doing so.