Perceptions and Sensors

We will use examples to get an idea of how information in the real world is perceived by sensors. Since we are dealing with robots, we use the word “perception” to describe the function, which is a little more human-like, but the word “perception” here is roughly synonymous with the words “measurement,” “measurement,” and “detection. However, the word “perception” here is roughly synonymous with the words “measurement” and “detection.

Sound perception

A sound sensor is used to perceive “sound”. The sound sensor is commonly called a microphone (microphone). The reality of sound is the vibration of air density (coarse and dense waves). When a light object is placed where the air is vibrating, that object will also vibrate. When the object is a magnet or metal, the vibration of the object is converted into an electrical signal. The rest of the process of perceiving sound is completed by converting the vibration information into electronic data using an electric circuit (electronic circuit or analog circuit).

Furthermore, the electronic data of the sound can be processed by a computer to extract information about the pitch and loudness of the sound. The computer can then perform speech recognition processing to identify the event that produced the sound and recognize it as a human word.

Perception of visual information (images)

It is the “image sensor” that is used to perceive visual information, or images. Since visual information consists of “light,” it is used to detect light. It is the “light sensor” that detects the light. The image sensor is made up of a large number of light sensors laid out in the form of pixels. For example, in the case of the fullHD standard, there are 1920 x 1080 pixels (about 2 million pixels). A digital camera is a product that combines an image sensor with a lens, LCD screen, button switches, and other components. Image sensors are also used in smartphones.

Each individual light sensor in each pixel is a sensor device called a “photodiode. Due to a physical phenomenon called the photoelectric effect, when light hits the photodiode, it generates an electrical charge corresponding to the amount of light. An electric circuit converts the amount of charge into electronic data, and when the information from all the pixels is collected and stored as image data, the perception of visual information is complete.

When acquiring image data, the time that light is applied to the photodiode (exposure time) is determined. This time is determined by the shutter. Light enters the photodiode only when the shutter is open.

Perception of the posture of the robot body

For example, to perceive the posture of a bipedal humanoid robot, such as whether it is falling or tumbling, “acceleration sensors” and “gyro sensors (angular velocity sensors)” are used. If the robot suddenly makes an unnatural movement, acceleration is generated, so if an acceleration sensor is attached to the robot body, unnatural movements can be recognized.

An example of a series of functions using an accelerometer is as follows. Set some threshold (standard) for acceleration. If the signal output from the accelerometer exceeds that value, the robot will recognize that there is a problem. If the signal from the acceleration sensor exceeds the threshold value, the system recognizes that something is wrong and activates a risk-avoidance function such as stopping the next movement of the robot.

On the other hand, the accelerometer alone cannot fully grasp the posture of the robot. If the robot “rotates” with the accelerometer attached to the center of gravity of the robot, the acceleration is zero, so we cannot know the rotation. To know the rotation, use the gyro sensor (angular rate sensor). By using both the accelerometer and the gyroscope at the same time, the posture of the robot can be completely known.

In general, to recognize the posture and motion of an object, we use a total of six pieces of information: acceleration information in the X-, Y-, and Z-axes and angular velocity information of rotation around the X-, Y-, and Z-axes in three-dimensional space. This information can be represented as vectors in a six-dimensional parameter space.

The above is a rough outline of perception and sensors. I will organize the details in a separate page.

Copied title and URL