"Is this reliable?" Chu Yuanxi asked while looking at it. It would be fine if he didn't ask. Su Gu printed things so professional that he couldn't understand. This technology is said to use Beidou navigation and does not use GPS, but Chu Yuanxi's world uses GPS for doing well-off life, so he really doesn't understand it.

Su Gu felt a little embarrassed. He just did this and didn't know before: "Well, I haven't used the Beidou system to develop shared bicycles before. I'm not familiar with it. I didn't expect to find it. Now Beidou is so powerful. I have spent time researching it these days, and there should be no problem."

"What's the principle?" Although Chu Yuanxi made a request himself, he didn't know how to solve it. With the technical characteristics of GPS back then, bicycles could be blocked by entering confined spaces, and even under the shade of the overpass, even under the shade of trees, positioning became a problem. Then something new to solve this stubborn problem.

The well-off in the original world did not realize that there was such a problem at the beginning, so the development stage did not put forward any requirements for accuracy. When the online content was made, it was very passive to find that there was no precise riding data support, so it relied on software to circle it, and the foundation was very fragile. So Chu Yuanxi made the request this time.

Su Gu picked up a piece of printing paper and said: "The principle is that the positioning chip of the bicycle is combined with Beidou's foundation enhancement station data to correct the satellite signal error."

"Isn't that the base station plan?" Chu Yuanxi was confused. Before, bicycles relied on base stations to locate when they couldn't count on GPS, so they were not sure. How could this achieve accurate positioning if the soup or the medicine was changed?

"Ah, because Beidou has achieved short messages." Su Gu understood this slightly, which was better than Chu Yuanxi. While reading the information, he said: "GPS can only transmit one-way signals from satellites to bicycle chips. Everyone knows that Beidou can realize bidirectional signal transmission of bicycle chips and systems through satellites and ground base stations, and the efficiency is much higher than GPS. When the GPS signal is blocked, Beidou terminals can communicate urgently through short messages, which can handle complex urban environments very accurately and minimize the drift area."

"Damn it's so awesome!" Chu Yuanxi picked up another promotional material. He couldn't understand the technical ones, but he could still read them. He found that his understanding of certain things was fixed in the state before time travel, and now he urgently needed to upgrade.

Moreover, upgrading is urgent. Technology is a thing that is a year away from being apart in this era. He broke his head when he traveled through time and could not even imagine that 5g was about to be commercially available. Do you dare to think about it at the beginning of 2018?

In fact, the development of Beidou is much sharper than Chu Yuanxi imagined.

First of all, in terms of chips, the domestic Beidou dual-band So and third-generation chips were launched one after another last year, which quickly made the market commercialized. The sales of compatible chips and modules exceeded 70 million pieces last year, and it was extremely cost-effective. If the bicycle is equipped with such chips with expertise in positioning, the positioning effect will obviously exceed that of general-purpose chips.

The so-called commercialization means that the hardware is hard enough and the variety of terminal products is rich, which can not only meet professional needs, but also face the public. Beidou is not just a chip, but also the sales of domestic high-precision boards and antennas have accounted for a large share of the domestic market. This makes it very easy for solution providers to take shape, and operation and maintenance service providers have emerged to provide industry applications.

This is the background of the emergence of "qxrtk technology", and not just shared bicycles, Beidou has begun to explode in other fields of the civil market, including automobile network, smart cities, portable equipment, and application terminals in various industries and types, and has launched a comprehensive stranglehold on GPS.

Beidou's next step is to form a global network, promote services overseas, and use technology and basic services to lose GPS's nest!

Chu Yuanxi thought, "I've been working on self-media for more than half a year, which is equivalent to the effort of closing my eyes and opening my eyes. I've actually gone so far from technology to market! Then he thought, "God, earth, finally, I don't have to use stupid gps anymore!"

This is actually a technical problem that Chu Yuanxi is most likely to hinder the development of Xiaokang's online content. According to the thinking logic in the original world, he is still preparing to mobilize resources to overcome difficulties. Unexpectedly, he has basically recovered without any medicine. The rest of the work is nothing more than Su Gu's search for the most suitable resources on the market.

Unexpectedly, when Xue Ming came out of Su Gu, he suddenly reported another good news. The reason why the popularity of coming out of troubled times is worse than that in China is that the translation of random plots is a big problem. The technical level of artificial intelligence translation is reduced to the player's experience.

There are two solutions. One is to use a large number of samples from two languages ​​to train Xue Ming's AI for deep learning. There are insurmountable difficulties, and the other is to improve the quality of AI.

The problem is that the translation is a bit cost-effective for Baren. The one I connected to before was already the best AI translation suite in China. It is likely that Xue Ming would train an AI to translate by himself and it is not as good as others.

So it's left alone.

But in February just past, Openai in Silicon Valley released a satellite and released a language-based general-purpose AI, which uses language modeling technology, can edit, translate and summarize the abstract, and is called gpt2. The reason is 2 is that the GPT generation has been launched long ago, using the same principle, but its performance is average.

The second generation project is not the case. It greatly expanded the training parameter set to reach an astonishing 1.5 billion parameters, and adjusted the training method of AI to make the goal simple prediction, summarized the previous words in the text, predict the next word, and generate an article.

This design idea is extraordinary, and the method is not a human way of learning, but the effect is surprisingly good. Because the main idea after adjustment is "inductive prediction", it can be violently swept all linguistic-related fields without any targeted training.

For example, when it comes to translation, the most difficult thing to grasp in translation is context rather than grammar and sentence structure. However, this AI does not need to learn the context, but directly summarizes the previous context through algorithms, and then "predicts" the correct words and sentences below. Of course, if used as a translation AI, a small output conversion verification model is required, which is slightly more complicated than writing and summarizing.

"You mean, can we use it? Then foreign players can see the original random dialogue?"

"Uh... it's not OK. You can only use a castrated version with 100 million parameters, but the effect is definitely better than the current one." Xue Ming said awkwardly: "Openai is not fully open source this time."
Chapter completed!