Human-computer interaction is the study of information exchange between human and computer, which mainly includes two parts of information exchange from human to computer and from computer to human, and is an important peripheral technology in the field of artificial intelligence. Human-computer interaction is a comprehensive discipline closely related to cognitive psychology, ergonomics, multimedia technology, virtual reality technology and so on. Traditionally, information exchange between human and computer mainly relies on interactive devices, which mainly include input devices such as keyboard, mouse, joystick, data costume, eye tracker, position tracker, data glove, pressure pen, and output devices such as printer, plotter, monitor, helmet display, and speaker. In addition to the traditional basic interaction and graphic interaction, human-computer interaction technologies also include voice interaction, emotional interaction, physical interaction and brain-computer interaction, etc. The last four typical interaction means closely related to artificial intelligence are introduced below.
(1) Voice Interaction
Speech interaction is an efficient interaction method, which is a comprehensive technology for human to interact with computer by natural speech or machine synthesized speech, combining the knowledge of linguistics, psychology, engineering and computer technology. It combines the knowledge from the fields of linguistics, psychology, engineering and computer technology. Speech interaction involves not only the study of speech recognition and speech synthesis, but also the study of human interaction mechanism and behavior in the speech channel. The process of speech interaction consists of four parts: speech acquisition, speech recognition, semantic understanding, and speech synthesis. Speech acquisition completes the recording, sampling and coding of audio; speech recognition completes the conversion of speech information to machine-recognizable text information; semantic understanding completes the corresponding operations based on the text characters or commands converted by speech recognition; and speech synthesis completes the conversion of text information to sound information. As the most natural and convenient means for human communication and information acquisition, voice interaction has more advantages than other interaction methods and can bring fundamental changes to human-computer interaction, which is the high point of future development in the era of big data and cognitive computing and has broad development prospects and application prospects.
(2) Emotional interaction
Emotion is a kind of high-level information transmission, and emotional interaction is an interaction state that conveys emotions when expressing functions and information, and evokes people’s memories or inner feelings. Traditional human-computer interaction cannot understand and adapt to human emotions or states of mind, and lacks the ability to understand and express emotions, making it difficult for computers to have human-like intelligence and achieve true harmony and naturalness through human-computer interaction. Emotional interaction is to give computers the human-like ability to observe, understand, and generate various emotions, so that they can eventually interact naturally, intimately, and vividly like humans. Emotional interaction has become a hot direction in the field of artificial intelligence, aiming to make human-computer interaction more natural. At present, there are still many technical challenges in processing emotional interaction information, describing emotions, acquiring and processing emotional data, and expressing emotions.
(3) Somatosensory Interaction
Somatosensory interaction is a natural interaction between an individual and the surrounding digital devices and environment directly through body movements without any complex control system and based on somatosensory technology. There are three main categories of somatosensory technologies, depending on the method and principle of somatosensory interaction: inertial, optical, and combined optical sensing. Somatosensory interaction is usually supported by a series of technologies such as motion tracking, gesture recognition, motion capture, and facial expression recognition. Compared with other means of interaction, somatosensory interaction technology has been greatly improved in terms of both hardware and software, and the interaction devices have developed towards miniaturization, portability and ease of use, which greatly reduces the constraints on users and makes the interaction process more natural. At present, somatosensory interaction is widely used in such fields as games and entertainment, medical assistance and rehabilitation, automatic 3D modeling, shopping assistance, and eye-tracking devices.
(4) Brain-Computer Interaction
Brain-computer interaction, also known as brain-computer interface, refers to the pathway that directly transmits information between the brain and the outside world without relying on peripheral nerve and muscle channels. The brain-computer interface system detects the activity of the central nervous system and translates it into artificial output commands that can replace, repair, enhance, supplement, or improve the normal output of the central nervous system, thus changing the interaction between the central nervous system and the internal and external environment. Brain-machine interaction transforms brain signals into machine commands by decoding neural signals, which generally includes three modules: signal acquisition, feature extraction, and command output. From the perspective of EEG signal acquisition, brain-computer interfaces are generally classified into two major categories: invasive and non-invasive. In addition, there are other common classifications of brain-machine interfaces: brain-to-machine, machine-to-brain and bi-directional brain-machine interfaces according to the direction of signal transmission; spontaneous brain-machine interfaces and evoked brain-machine interfaces according to the type of signal generation; brain-machine interfaces based on EEG, brain-machine interfaces based on functional MRI and brain-machine interfaces based on NIR spectral analysis according to the signal source.