Carben's dilemma is a natural state of industry technology moving forward, the pathfinder is always harder than the walker, and in the same way, if the pathfinder finds a new path, he has the opportunity to reap the greatest value. Pen × fun × Pavilion www. ｂｉｑｕｇｅ。 ｉｎｆｏ

In the concept of Mo Hui, the obstacles faced by Carbon are basically engineering and technical obstacles, and there are very few theoretical obstacles, and at this stage of human development, life service robots have actually matured in various major basic theoretical fields, which is nothing more than a large number of problems in engineering technology.

To make a simple analogy, the theoretical model of the internal combustion engine has been established for a long time, but the successive generations of internal combustion engines have been continuously updated and the technology has been continuously advanced, in fact, it is not a theoretical breakthrough, but the engineering technology of the internal combustion engine is improving day by day.

If we put aside the progress of peripheral disciplines such as material technology and talk about the technological progress of the internal combustion engine alone, we will find that its progress has not deviated from the initial theoretical model in principle, but is nothing more than the continuous improvement of thermal efficiency and power, which are all engineering and technological progress.

Now in the field of robotics is facing this problem, the main field of theoretical technology is no longer an obstacle, now just need to make a breakthrough in engineering technology, theoretically feasible, but also must be in the engineering to achieve it.

At present, several key areas of robot jamming, image recognition, speech recognition, artificial intelligence, positioning and navigation, to be precise, are not stuck, but the existing technology is not well realized.

Just like the early steam engine, the pressure is not good, the seal is not good, the transmission is not good, and the mechanical structure is not good, resulting in the overall efficiency is very low, and can only be responsible for drainage in the mine, and the application scenario and market acceptance are very limited.

The current robot is also in this state, on the whole, there are technologies in each field that can be used, but the performance is not good, the combination of the whole is worse, often quite expensive things, but the real use is a moment of freshness, applicability and work efficiency are very poor.

To put it bluntly, there are too many areas that need to be strengthened in the current life service robots, and the technology in these fields is too low, resulting in the overall application performance of the robot can not be improved.

However, there is an advantage that all relevant technologies are available, so it is not that there is a technical blank area that cannot be solved completely for the time being, and there is no problem that has been solved, and now the problem of whether it is good or not is being solved.

For example, image recognition technology, this technology has been around for a long time, and many related application technologies have been extended from this technology, such as Baidu's search map, such as face recognition, such as three-dimensional reconstruction, etc., all of which are extended from this technology.

KUKA is facing standardized designable scenarios, while Kaben is facing random and uncontrollable scenarios, and there are many emergencies, so relatively speaking, the technical difficulty faced by Kaben is much higher than that of KUKA. It's just that KUKA tends to be precise and efficient, and Kaben is inclined to usability and intelligence.

The acquisition cost of Carben is not high, and it is the furthest in the field of life service robots, one of the main considerations for Mo Hui to win Carben is to try to use ultrabooks to accelerate the research and development process.

At present, the computing power of the ultrabook is extraordinary, and the intelligence is not bad, although it may not be the strongest, at least it is the first echelon in the AI developed by major laboratories.

The possible help of ultrabooks for Kaben should be in simulation and troubleshooting, such as image recognition, and the key to obtaining an image recognition technology with good application effect lies in the intelligence of algorithms and unsupervised self-learning.

Mo Hui took a stupid approach, and when Carbon was acquired, he asked the Carbon team to provide an evolutionary model in the field of image recognition, which is actually a training model for artificial intelligence training and learning through massive samples. At the same time, let them provide all possible algorithms in the field of image recognition, regardless of the effectiveness of this algorithm.

Mo Hui had quite a lot of resources at his disposal by the Hand of God, and he not only asked Kaben for it, but also contacted many scientific research and commercial institutions that were conducting research in this field and asked them to provide similar things.

What Mo Hui does is actually try to exhaust the algorithms in this field, and at the same time exhaust the research methods in this field, and then use a large number of samples to make the ultrabook infinitely compare and combine.

Strictly speaking, this is not scientific research, he is using the computing power of ultrabooks, constantly permuting and combining, exhausting all possibilities, and finding a possible path in it.

Although this method is very stupid, it can really be effective in a certain type of field, in fact, all kinds of laws and correlations extracted from big data are using various algorithms to find hidden contingencies in a similar way. Theoretically, some scientific research is also similar to exhaustive, and the invention of incandescent lamps is actually an exhaustive list of all possible materials, and finally tungsten was chosen.

The first scientific research of the ultrabook was carried out with the help of Azhu, an artificial intelligence, but fortunately, the computing speed of the ultrabook is very fast, and the technology will give results instantly, no matter how large the sample pool is.

Therefore, Mo Hui's scientific research progress is very fast, and he can only adjust the calculation model dozens of times a day, constantly trying and making mistakes and constantly colliding, looking for the possible right path.

With image recognition technology as a whetstone, Mo Hui constantly sharpened the application methods of ultrabook scientific research, constantly adjusted, and constantly tried, and in the continuous interactive adjustment, ultrabook slowly brought its super performance into play.

When the scientific research method of ultrabooks has basically taken shape, an image recognition with unsatisfactory application performance has been released. Its algorithm is the result of countless extravaganza and evolution, and its smart module has been trained on hundreds of millions of images and videos to be adaptable enough.

This image recognition technology was quickly tested by the Carbon team, and it was found that it was able to recognize most common items in the home through photos and videos, and it was able to distinguish pets and moving objects.

To do this,In fact, in the visual recognition module,It can almost meet the needs of life robots,What needs to be done later is based on this technology,Extend other functions,Such as distance judgment,Path planning,Self-positioning and so on。

For Mo Hui, the biggest gain is not this image recognition technology, but the scientific research mode and method explored by Ultrabook in the process of developing this technology.

With the first step, Mo Hui programmed it into a new module with self-learning ability and quite intelligence - Scientific Research 1.0 (to be continued). )