Gradually into the family of intelligent robots

"Companionship is the best confession." Sci-fi movies of intelligent robots, already meet the spiritual needs of mankind. Because more intelligent than pets, future robots should be the best playmate. So, at this stage the robot to what extent? Humans should look at what these robots look at it?

In a recent Suzhou mems (MEMS) market at the annual meeting, the domestic home robot experts to apply the requirements of the sensor as a context, carding a robot to accompany the development of ideas.

Home robots are generally divided into three categories

If you can, you will purchase a home, what kind of robot?

Currently, the family of intelligent robot can be divided into three types, the first is the application of robots, has a very specific function, such as the increasing popularity began sweeping robot, window cleaning robot. The second is accompanied by social robots, and human interaction can accompany users grow together. Many domestic and foreign enterprises have begun to accompany the development of social robots, robot such as Alpha, excellent small, and in the next one to two years, accompanied by family social robots could become the focus of the consumer market.

The last category is the bionic robot, bionic creatures from the principle of movement, expression, and other ways of thinking, such as Honda R & D asimo robot can perfectly imitate the movement of the feet, to complete a lot of very complex movements, including up and down the stairs, the United States emotional robot Sofia feedback can be based on different expressions of human language, and to achieve the purpose of human communication.

The brain is the cognitive processor core technology

"Whether it is the application of robots, robot social companionship, or bionic robot, the core functional requirements are similar, namely, perception, cognition and movement behavior control." Where the beans Dongguan, general manager of Information Technology Co., Ltd., said Zhu Xiaoming , is perceived by mems sensors, sensing the external environment; cognitive robot brain is the information analysis and processing, and then converted to the experience of the robot so that it can grow along with experience; motion control behavior can be understood as robot control body movements, Wink feedback.

The brain is the processor core cognitive technology, determines the degree of intelligence robot. Designed as ibm supercomputer watson, beat human players in the US television quiz shows in jeopardy and famous, caused quite a stir. Watson then be applied in many fields of medical, banking, and other investigative bodies, and ibm watson is trying to make more intelligent than humans.

However, at this stage the market to see most of the social companionship type robot, still using the traditional mobile phones or tablet proven solutions, although to achieve the basic functions, but not targeted. Because the different types of robots, emphasis should be selected or a different function.

"Social robot should have accompanied the cloud the brain, otherwise only have limited knowledge of toys, would not be accompanied by the user's smart growth and growth." Zhu Xiaoming said, "but not all of its intelligence capabilities through the cloud to realize the need in the cloud local and make a reasonable distinction to ensure that in the case of disconnection, the robot is still able to complete the basic work. at the same time, data processing, too, need to optimize both local and cloud storage and processing. "

Robot perception and feeling

Robotics and mems industry are closely linked, because all can think of mems sensors, robots are reflected, to promote the development of intelligent robots. Let's look at several major technology status.

Voice is the most basic requirements of the robot, from the initial story machine point reading pen to function gradually rich cloud robot, there are also many in the sale of voice interactive robot, robot pocket as language, music as music Andy launch, children's education excellent small robot, an overseas cogni-toys, with its core values through voice architecture, implementation, and child interaction.

Get voice through a microphone, and then extracts and saves the user's voice characteristics, through the terminal and cloud complexes, for some simple voice, the terminal can be implemented directly. However, for complex speech, voice robot ships will be sent to the cloud to identify, and then return to the terminal, then you might see a noticeable pause exchange.

In addition, the cloud of light energy to understand language is not enough, we must further understand the semantics, robot brain truly understand the user's semantics. Therefore, the robot through semantic understand and grasp the needs of users, and user feedback and can verify the accuracy of semantic understanding, and thus gradually develop intelligence IQ robot. Because the language is very rich, so the semantic understanding of the ever-changing will need to go through a long period of accumulation statements and manner of speaking, by the accumulation of large amounts of data to analyze semantics. At the same time, we still need to identify the feelings, the user becomes higher tone, speed becomes faster when the user can understand the anger is in the state. Stage really totally dependent voice control application is still relatively small. In addition, in a complex environment, recognizable sound itself will decline, speech recognition will also be more difficult.

In the family, the robot can remember the best way for each family member, identify the user's facial expressions, gestures and even recognition, intrusion of strangers. Accordingly, vision is also essential to the function. This is the face recognition. Face recognition feature to find the point of people face, the robot can also reach further identify emotional effects, such as: eyelid movement, chin up or down, how about lips, etc., you can know the meaning behind this expression. In addition, gesture recognition and gesture recognition, can be captured by the camera, the robot more intelligent.

Haptic system will make the robot more intelligent

"Whether it is face recognition, gesture recognition or gesture recognition, in fact, is the essence of image processing and understanding." Zhu Xiaoming, said chip solution architecture of the robot can be optimized from the current mobile phone cpu + gpu chip architecture to start, according to the specified cpu do recognition, data processing, and operating system, GPU division can handle a large amount of image information, thereby improving the efficiency of re-optimization architecture.

In addition, the touch sensing allows robots and human interaction easier, "funny look, or give a robot scratching, or slap greeting, robots can feel, and make the right feedback. If you add artificial skin real texture, or can the external surface texture and physical properties to achieve a comprehensive perception, such a robot haptic system will allow them to do anything more. "

Two factors research and design of social robots accompany the most important consideration, one functional goals need to be clear that what the robot is mainly to meet the needs of users; second is the development of the cloud, which is cognitive, the robot can more clever, can help users solve their key problems.