Configuration of Robots

This section summarizes the composition and general story of robots from the author’s dogmatic point of view. As described in the definition of a robot page, a robot includes the functions of “perception,” “recognition,” “judgment,” and “action. The hardware corresponding to these functions are “sensors”, “computers”, and “actuators (drives)”. Sensors are responsible for perception, computers for recognition and judgment, and actuators for motion. A robot is a “system” that combines these elements. (A system is a combination of multiple elements to achieve a single function.)

Configuration of Robots

There are some areas where this relationship does not strictly hold, but we will think of it simply to make it easier to organize the technology.

As an example of a part of the system that does not support this, let’s take a fan heater system that wisely controls the system according to the dwelling environment. Suppose the fan heater is equipped with a temperature sensor to recognize the living environment, and the built-in microcomputer (computer) processes the temperature information obtained from the sensor to adjust the intensity of the heater’s warm air in a nice way.

In some of these systems (products), the temperature sensor module has a built-in function that outputs an electrical signal that forcibly turns off the hot air of the heater when the temperature reaches a certain standard (e.g. 40°). This is a function to avoid danger in case the computer does not work properly. It is “recognition” that compares whether the temperature is higher or lower than the reference temperature, and “judgment” that decides to turn off the heater based on the result, but this is done not by the computer but by a basic electric circuit (if it is just comparison, it is done by a circuit called comparator).

This electrical circuit is included in the sensor module, and the “recognition” and “decision” are executed in the “sensor”, which is outside the aforementioned correspondence.


A sensor is a device that converts a physical quantity (physical phenomenon) into electronic data. A sensor is a device that converts a physical quantity (physical phenomenon) into electronic data. (if the term “physical quantity” or “physical phenomenon” sounds too formal, you can replace it with “the real world that humans can sense.) For example, an “optical sensor” converts the intensity of light into electronic data.

If the object is sound, there is a “microphone (sound sensor),” and if it is the movement of an object, there is an “accelerometer” or “gyroscope. In a narrow sense, a sensor is a device that converts a physical quantity into an electric current or voltage that can be handled by an electric circuit (analog circuit), but here we call it a sensor including the part that converts it into electronic data (digital information) that can be handled by a computer.


For the purposes of this website, we will broadly classify devices that realize the functions of judgment as “computers. When we say “computer,” we generally mean a personal computer or server machine, which functions with the CPU (Central Processing Unit) at its core.

Recently, GPUs (Graphical Processing Units) are also often used to efficiently perform “deep learning” and “inference operations with neural networks” created by deep learning.

I believe that GPUs are also becoming generally recognized as a type of computer. In addition, FPGAs (Filed Programable Gate Array), DSPs (Digital Signal Processors), and LSIs (semiconductor integrated circuits) dedicated to artificial intelligence are sometimes used to realize the functions of artificial intelligence and robot decision making. I would like to include these devices in computers.

In many cases, the “recognition” and “decision” functions of a robot are expressed in the form of “software” and incorporated into a computer (product). The software includes elements such as artificial intelligence programs, non-artificial intelligence programs, and operating systems (OS).

Artificial intelligence includes, for example, neural networks created by machine learning (e.g., deep learning), command programs for robot movements created by reinforcement learning, and programs that intelligently classify information obtained from sensors (e.g., support vector machines).

Programs that are not artificial intelligence include basic programs such as starting an operation when a product switch is turned on, or instructing a button to change the intensity of the heater’s warm air so that it is directly reflected in the heating part of the heater (the actuator part).

Also, when control with “feedback” is used, mathematical expressions for control operations (PID, etc.) are sometimes included in the program. The OS for the last element mentioned refers to things like Windows and Linux.


Actuators (drives) are devices that convert electronic data into physical quantities (physical phenomena). They reflect the results of computer calculations in the real world that humans can feel. In general, the term “actuator” is often used to refer to a motor or other device that makes a visible movement.

On the other hand, some communication robots only emit sound or light. Since we would like to treat these as robots as well, we would like to define actuators broadly, not limiting them to those that perform mechanical actions. Actuators will include motors, light sources, speakers, heaters, etc.

In English, a device that outputs sound is sometimes called a “sound actuator” and a device that outputs light is sometimes called a “light actuator,” so including devices that emit sound or light in the definition does not seem to deviate much from the original meaning of actuator.

There is also the term “output device,” but this term strongly reminds me of the output device of a “PC” (LCD display or speakers), so I cannot say that this is appropriate. In this site, I will use the word “actuator” at my own discretion (preference).

Infrastructural elements

The actual fabrication and manufacturing of a robot cannot be completed with only the components of sensors, computers, and actuators. All of these devices need “power” to operate, so a “power source” is necessary. There are three main types of power sources: a cable that connects to an electrical outlet, a battery, and a power generator.

In addition, you will need a box to hold the sensors and computer (called a “chassis”), and “materials” to give shape to the actuator mechanism. Some robots also require a “network” to communicate with the outside world.

Although these elements do not carry the direct functions of the robot, they play a role as the minimum necessary “infrastructure” to make the robot work. Therefore, we call these elements the “infrastructure” of the robot system.

At first glance, it may seem like a supplementary element, but it is an important element that affects the overall performance of the robot, as the amount of energy stored in the battery, for example, can significantly change the amount of time the robot can operate.

The above configuration of the robot is illustrated in the following image.

Configuration of Robots

Copied title and URL