Shulou(Shulou.com)06/02 Report--

As a map user, recalling the surprise and excitement of finding his birthplace on Google satellite maps, I still lament the shock that technology brings to every ordinary person-- it only takes a computer to have a "God's perspective". You can travel around the world with a swipe of the mouse.

But the last time I spent time browsing world-famous scenic spots and streets through Google Satellite Maps and Street View Maps was more than a decade ago.

Yes, Google Maps has been out of sight for ten years. But if you pay a little attention, you still need to know that Google Maps is still the super APP that affects mobile users around the world. Google Maps has more than 1 billion monthly active users; using it, people travel 1 billion kilometers a day, producing more than 20 million comments and ratings.

For Alphabet, Google Maps is still the core product besides search, Android, Google Store and YouTube. Analysts predict that Alphabet's advertising revenue from Google Maps is likely to reach $3 billion-$4 billion in 2018 and is still likely to grow at an annual rate of 25-30 per cent over the next three years.

Google Maps celebrated its 15th birthday on February 8 this year. For this anniversary, Google Maps has made a major revision and upgrade on iOS and Android. Meanwhile, Alphabet and Google CEO Sander Pichai posted a blog post saying that Google Maps will be an important position for the company's AI First strategy in the future.

Sandel's statement seems to convey two layers of message: the first layer is that in the five years since Google announced the launch of "AI First," Google Maps has not become an important front for AI technology support; the second layer means that Google Maps will focus on the application of AI technology in the future.

Since Google announced its full embrace of AI in 2016, the entire range of Google products have been upgraded through machine learning and deep learning technology. Compared with the star products such as Waymo, Google Assistant, Gmail, Google Translation and so on, Google Maps does seem a little flat and slow in the application of AI technology.

Being slow doesn't mean doing nothing. As a Google Maps covering 1 billion users around the world, the AI of its products has already been imperceptibly carried out. Being insipid doesn't mean it's not worth paying attention to. The idea of AI-based Google products is to gradually infiltrate AI technology into its products to enhance the user experience.

In the field of science and technology, we always overestimate what can be done in a year or two and underestimate what can be done in five or ten years. This is true for search engines, so is self-driving, and so is Google Maps. In the long period of five years, some of the technology depth of Google Maps is worthy of our attention again.

Google Maps, born AI?

Time goes back to 2016. Sandal Pichai, then CEO of Google, announced on Google Event2016 that Google's overall strategy had shifted from "Mobile First" to "AI first". As we all know, Google is well prepared for the transition to AI strategy. Google began to lay out machine learning as early as 2010; in 2012, it released a "knowledge graph" and built a super-large neural network system, "Google brain"; and in the next two years, it included deep learning company DNN Research and British artificial intelligence leader DeepMind; then, Google AI results like a blowout, 2015 Google open source deep learning framework TensorFlow, and gradually applied to more than 50 Google products In 2016, AlphaGo, developed by DeepMind, was born, bringing AI into public view.

We know that the development of AI needs the support of algorithm, computing power and big data. Among them, the map itself has a natural fit with AI.

On the one hand, the map product is the mapping of the real world, and many of people's offline movements will be reflected through this product and accumulated into data. The same is true on Google Maps, whether it is billions of monthly user travel records and uploaded location tagging data, or PB-level satellite map data and global street view image data, it can be called big data's "rich mine".

On the other hand, the function of the map itself can also be used by a large number of AI-related industries, such as smart cities can not do without the use of map products for thermal tracking of traffic flow, autopilot also needs to combine high-precision maps with radar sensors.

So for Google, which dabbles extensively in AI, maps are naturally a good testing ground for technology.

The most typical of these is the use of Google Street View data. Google Street View was originally a feature of Google Maps, which was shot by a dedicated street view car, and then put 360-degree real-world photos on Google Maps for users to use. Soon when Google came into contact with AI, especially when Li Feifei, who had trained on imageNet, joined the job, these huge amounts of data from real-world images gave them a chance to show their talents.

For example, using an enhanced deep neural network, thousands of street view images in California are scanned and converted into professional photos. For example, the application of face recognition, OCR recognition functions, the street view image of the license plate, face code to achieve the purpose of privacy protection. In the same way, roadside street numbers, business names, speed limit signs and other details are also being extracted from the pictures, and new addresses are automatically created and located on Google Maps as appropriate. Similarly, the model can also be applied to the name recognition on the exterior wall of merchants. Through this feature, you can also update the changes of merchants more accurately and continuously.

In other words, AI has been optimizing the underlying technology of Google Maps.

(the system automatically recognizes the merchant as "Zelina Pneus" without the information of the real address.)

In addition, Google Maps demonstrated its AR-based Walk Navigation system (VPS) for the first time at the Google Imax O 2018 developer conference. The AR system actually uses Google Lens products based on image recognition and OCR technology, which can be used to navigate through mobile cameras to capture and identify landmarks. Just click the "start AR" button and Google Maps will show users a video image in which the forward arrow overlaps the camera. Behind the flexible and simple AR navigation at the front end, a large amount of street view data is needed at the back end; if the exact location of the user cannot be obtained through GPS, Live View will also use machine learning to compare the scene captured by the user's camera, and then reposition it with billions of street view images.

In addition to images, Google also uses big data's recommendation and prediction to provide a direct user-oriented service experience for Google Maps AI. Google Maps offers AI personalized recommendations for commuting efficiency optimization and mixed travel modes. For example, according to the real-time calculation and prediction of each real-time road condition and bus information, you can help commuters tailor each commuter route; for non-self-driving mixed commuting mode, Google Maps will make a comprehensive calculation based on the walking time, cycling time and bus arrival time, and give commuter advice for users to make decisions.

In addition to intelligently recommending routes, Google Maps adds more merchants and new addresses to the map through the combination of AI and satellite images. It can also display the opening and closing time of the target store selected by the user, and the average length of customer stay in the store, so as to remind the user to leave at the right time, so as to avoid not finding parking spaces or encountering nearby road congestion. This is closely related to the multi-dimensional data provided by users and merchants collected by Google Maps for a long time. Although these new AI functions are considered by many people in the industry to follow in the footsteps of similar APP in China, there is a lack of innovation. But the details recommended by the AI algorithm are very noteworthy.

In 2018, for example, Google Maps launched real-time bus delay forecasts based on machine learning in hundreds of cities around the world. In addition to relying on real-time data on travel time and bus location provided by public transport agencies, the model created by Google Maps also combines factors such as speed, time, stops and road congestion to make real-time predictions. Combined with travel data and machine learning models, they can also predict bus or train congestion in 200 cities around the world.

This detail is very practical. When you are still a few hundred meters from the bus stop, maybe Google Maps will tell you whether it is worth taking a quick walk to catch a bus that will be delayed for a few minutes. When you learn that there is an empty car with seats right behind you, will you give up the overcrowded bus in front of you?

In addition, the system is based on machine learning. By comparing a large amount of data on users' personal preferences, the system can customize personalized information for different users, such as telling you what new stores have opened near where you live and recommending nearby food to you. Users can also quickly share the delicious food on the map with their friends and mark it in real time.

Based on the data, the "calm" AI upgrade road

As can be seen above, the combination of massive data such as image data and user behavior data has become a powerful moat for Google Maps to remain standing. With huge amounts of data, the AI upgrade of Google Maps presents some more obvious features.

The first is the heavy reliance on Google's graphics technology. From the above AI evolution path of Google Maps, Google is the first to apply machine learning and other technologies to deal with huge map image data, including the recognition of important information such as road traffic signs, street names and store names. On the one hand, it stems from Google's own leading edge in image recognition technology, and on the other hand, it is also Google's priority consideration for accurate travel map data.

The second is the in-depth mining of user travel data and local service data by Google Maps. Using AI algorithm, a large amount of data hidden under users' local search can produce better travel service recommendations and more business value. Google has found that in addition to travel navigation, exploring local services has also become a heavy use tool for users. At present, Google Maps "search" tool can satisfy users to book hotels, cars and meals, query travel routes, and even meet dozens of different types of services like Meituan in China.

For example, the update of the new version of Google Maps is mainly an update of the five tabs on the home page: explore, commute, save location, contribution and update (Explore,Commute,Saved,Contribute and Updates).

From these new features, it can be seen that Google Maps will pay more attention to users' localization services and users' active contribution to geographic data. With the perfect travel data and the recommendation of AI algorithm, Google Maps can respond to the needs of more and more users in real time and provide more travel services.

In addition, we can also see the ambition of Google Maps to integrate into global city management. Although we rarely hear that Google is involved in the so-called "smart city" project. But Google is already laying out this huge project in depth. For example, Google Maps has begun to use machine learning algorithms and satellite images to draw infrastructure such as complex buildings around the world. According to Google, algorithmic rendering alone added 110 million buildings in the first half of 2018. It can be predicted that what Google will give in the future is no longer a two-dimensional geographic schema, but more likely to be a three-dimensional 3D map world.

Through the above observation, Google's layout in AI technology pays more attention to the excavation and processing of big data at the bottom, and pays more attention to the improvement of service detail experience. This makes Google Maps almost "lost its voice" in technology news in recent years. As a product that has served Google AI for many years and coincides with the 15th anniversary of this node, there are no key updates on voice interaction, no news related to car networking and self-driving. Such detailed technical improvements will inevitably make people feel a little dispirited. It seems that the technology upgrade of Google Maps can also be upgraded to congested roads, and the speed is not satisfactory.

AI unsatisfied: Google Maps'"change and immutability"

Another obvious change in this version of Google Maps is the update of the icon. The new icon removes the classic map style and replaces it with a Google-toned map pin. Google's internal explanation for this change is not only to help users navigate, but also to help users find where they want to go.

In fact, although from a business point of view, the 15th anniversary of the technology update may not be sexy enough. But from another perspective, Google Maps is still doing its job: helping users achieve a better travel experience and provide better public service.

Google Maps, for example, is predicting bus delays in real time through machine learning. By predicting the delay of the bus in advance, even if it is only a few minutes of error, it can also be of great help to users. For example, enhance the data connection to public transport, allowing users to submit more detailed information about public transport, such as car temperature, wheelchair accessibility or whether there are female-only carriages, in order to make more user-friendly and considerate travel recommendations.

And some more complex public services. Combined with disaster weather, epidemic and other prediction models, predict and timely inform local users of some unexpected situations, in order to avoid hurricanes, tsunamis, earthquakes, infectious diseases and other unexpected disasters.

Finally, it is worth mentioning that an artist user used a "hacker" to commemorate the 15th anniversary of Google Maps.

The German artist, Simon Wechkert, towed a trolley with 99 mobile phones in navigation mode on the road. as a result, he successfully mistook Google Maps for nearly 100 cars on the road, causing heavy traffic jams throughout the road. To this end, he specially chose to experiment in front of Google's Berlin headquarters building before the anniversary celebration. Wechkert's obsession is to verify that the way Google Maps determines congestion is still very simple.

To this end, Google responded: this creative behavior is very encouraged, and we are willing to accept users' opinions to make Google Maps better.

This seems to be the official style of Google's version of "never forget the original ideal and ambition, keep in mind the mission". It is not known when this small problem will be solved on the congested road of technology upgrading.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.