(This article was published in "Communication World All Media", the author is Yang Xuecheng;)

　　These issues in the smart home industry are worth pondering:

① Smart home allows set-top boxes to enter the home, but there is no practical solution for the networking of electrical appliances;

② The meaning of "entry" and the question of "who enters" are worth rethinking;

③The core of intelligent hardware development cannot be separated from "intelligence", which must be accurate, predictable, and benign operation.

　　Smart home has been popular for a period of time three years ago, mainly due to the eyeball economy caused by Google's high price acquisition of Nest and the direction indicated by the Apple Home Kit development platform to the industry. But since then, there have been few major breakthroughs in the smart home field, and Nest's follow-up performance has not surprised the industry. This field seems to be a forgotten corner.

More than a millimeter: thinking about smart home

　　The fundamental reason for the lack of breakthrough in the smart home market lies in the complexity of the scene, that is, the practice of related manufacturers trying to use an "entrance" to occupy the "family" is often difficult for users to accept, coupled with the inherent interests of traditional home appliance manufacturers and standard standards. The difference has brought a very high user conversion cost.

　　Although products such as set-top boxes or set-bottom boxes have greatly changed home TVs, there is still a long way to go before the full transformation of all home appliances from electrical appliances to "network devices". There seems to be no practical solution to the networking of electrical appliances.

　　This situation forces us to rethink the meaning of “entry” and the question of “who enters”. In fact, the entrance is not in size, but in the richness of interaction. Although the physical area of ​​the TV screen entrance is large, it is still unable to compete with the small mobile phone screen at the entrance level because of the rich interactive nature of the mobile phone entrance.

　　Therefore, thinking about the entrance of smart homes should get rid of screen thinking, and should work hard on interactive methods. How to create diversified interactive methods through household electrical equipment is the fundamental, infrared, microwave, gesture, voice, image, touch, etc. are all possible Build a unique interactive scene.

　　No matter what kind of smart device, and no matter what kind of interaction method, the development of smart hardware is inseparable from the three major steps of "embedding-interaction-evolution".

　　First of all, how to seamlessly embed the scene is the test that smart hardware will face in the first step. Since there are physical products, users must undergo psychological construction to accept it, otherwise there will be a sense of imbalance, so the promotion and embedding of smart hardware is facing high The threshold is either to provide tangible technical added value, or to reduce the price to a very low level. Of course, being able to immerse the product in an existing scene is the best practice.

　　The second is interaction. The interaction must be two-way. There must be a channel to empower the terminal, and there must be a way to transmit data to the cloud. The two phases can be connected to succeed in the interaction.

　　The last is to lead the evolution of hardware. It should be known that the ecological construction of the digital economy follows the "layer barrier structure", which extends infinitely on the plane and continuously superimposes on the levels.

　　This layered structure makes the basic logic of traditional market segmentation not established in many cases. "The market has no segmentation, and the business layer is added" is the correct posture. Each layer is stacked, the product evolves and upgrades once, and the data is rich On the first floor, the result of the cycle is the formation of an ecosystem, which is somewhat similar to the working principle of a 3D printer.

　　Of course, the core of smart hardware lies in the word "smart". The so-called intelligence is essentially a change in the paradigm of human-computer interaction. In the past, the machine was a reactive responder to human instructions. You gave an instruction and the machine output feedback.

　　This stimulus-response system requires that the results must have "precision", must be accurate and predictable, so that the stimulus-response system can function well. And artificial intelligence is to turn the human-computer interaction method into a "feedforward active service system". No interaction is the best interaction. You don't need to give the machine any instructions, the machine will take the initiative to perform services, so the key behind it is no longer It is "precision", but "temperature". The artificial intelligence system must have temperature.

　　To achieve the temperature, big data and deep learning become extremely important. The data analysis supporting the intelligent hardware system is inseparable from three elements: computing power, algorithm, and computing efficiency. Computing power refers to the abundance of real-time data and data energy reserves. Algorithms are responsible for converting computing power into scenarios where users can perceive "temperature", and computing efficiency is the level of cost-benefit index. Only with the ability of self-learning at the bottom level can smart hardware products interact and evolve better after being embedded. This is a systematic project.