A typical robot system consists of four interacting parts: the manipulator, the environment, the task, and the controller. We call the sensors typically mounted on the robot's manipulator "inner sensors," and the sensors that are part of the environment "external sensors." The following discussion will primarily focus on this classification, while also incorporating other methods for classifying robot sensors.
Robot products are currently classified into two main categories: industrial robots and service robots. Domestically, some also categorize them into industrial robots and special-purpose robots; general robots and intelligent robots; general robots and mobile robots; or general robots and humanoid robots, etc. Currently, industrial robots are mostly used for handling, sorting, loading and unloading, packaging, palletizing, welding, spraying, grinding, polishing, cutting, placement, and assembly.
With the increasing level of intelligence, the application of robot sensors is becoming more and more common. Intelligent robots mainly fall into three categories: interactive robots, sensory robots, and autonomous robots. From the perspective of anthropomorphic functions, vision, force, and touch are the most important and have already entered the practical application stage. Hearing has also made significant progress. Other senses include smell, taste, and tactile sensation, each corresponding to a variety of sensors. Therefore, the robot sensing industry has also formed a production and research force.
A robot's control system is analogous to the human brain, its actuators to its limbs, and its sensors to its senses. Therefore, for robots to receive and process external information like humans, robot sensor technology is a crucial aspect of robot intelligence.
Sensors are essential for robots to perform sensory operations. Through their sensory function, sensors convert the robot's own characteristics or the characteristics of related objects into the information needed for the robot to perform a certain function. Depending on their purpose and scope of application in robots, sensors can be divided into internal sensors and external sensors.
Internal sensors are used to detect the robot's own state (such as the angle between the arms, the robot's position, speed, and acceleration during motion engineering); external sensors are used to detect the robot's external environment and the condition of objects, such as the shape and spatial position of the object being grasped, whether there are obstacles, and whether the object has slipped.
Main categories of robot sensors:
Internal sensor
Mechatronics products integrate sensors and mechanical components or structures such as motors, shafts, arms, and wrists to measure position, speed, and force, thereby achieving servo control.
Position (displacement) sensor
Linear motion sensors include potentiometer-type sensors and adjustable transformers. Angular displacement sensors include potentiometer-type, adjustable transformers (rotary transformers), and photoelectric encoders. Among these, photoelectric encoders include incremental encoders and absolute encoders. Incremental encoders are generally used for position servo control where the zero position is uncertain. Absolute encoders can obtain the instantaneous angle value of the drive shaft corresponding to the initial locked position of the encoder. When the device is subjected to pressure, by simply reading the encoder readings of each joint, the setpoint of the servo control can be adjusted to prevent excessively violent movements when the robot starts.
Speed and acceleration sensors
Speed sensors can measure both translational and rotational speeds, but in most cases, they are limited to measuring rotational speed. A photoelectric pulse speed sensor utilizes the derivative of displacement, particularly photoelectric methods, to illuminate a rotating disk and detect the rotational frequency and number of pulses to determine the rotational angle. Alternatively, by creating a slit in the disk and using two photodiodes to identify the angular velocity (rotational speed), this sensor can be described as a photoelectric pulse speed sensor.
In addition, there are tachogenerators used for speed measurement, etc.
A strain gauge, also known as a tensile testing instrument or stress sensor, is used for acceleration measurement. Accelerometers are used to measure the dynamic control signals of industrial robots. Common methods include extrapolation from velocity measurements, extrapolation from the force generated by the acceleration of a known mass object (using a strain gauge to measure this force), and the methods described below:
The force associated with the measured acceleration can be generated by a known mass. This force can be electromagnetic or electrodynamic, ultimately simplifying to a measurement of current. This is the servo return sensor, which can actually be further divided into various vibration-type acceleration sensors.
Force sensor
Force sensors are used to measure the three components of the force and the three components of the torque between two objects. An ideal sensor for robots is a semiconductor stress gauge bonded to a follower component. Specific types include resistive force sensors, semiconductor force sensors, other magnetic pressure sensors, and force sensors based on the principle of string vibration.
Other types include torque sensors (such as photoelectric sensors for measuring torque) and wrist force sensors (such as the Stanford Research Institute's sensor, which consists of six small differential transformers and can measure the force acting on the wrist in the X, Y, and Z directions, as well as the torque on each axis).
Because of the long history of robot development, in recent years, AC servo systems based on AC permanent magnet motors have been widely adopted. The corresponding sensors for position, speed, etc., are mostly made up of various types of photoelectric encoders, magnetic encoders, and rotary transformers.
External sensor
Traditional industrial robots generally lack external sensing capabilities, but new-generation robots, such as jointed robots, especially mobile robots and intelligent robots, are required to have the ability to correct and respond to changes in the environment. External sensors are used to achieve these capabilities.
tactile sensor
Miniature switches are the most common type of contact sensor. Other types include isolated dual-state contact sensors (i.e., bistable switching semiconductor circuits), single analog sensors, and matrix sensors (matrix sensors with piezoelectric elements, artificial skin—variable conductivity polymers, light-reflective tactile sensors, etc.).
stress sensor
When a multi-joint robot performs an action, it needs to know three conditions: the actual contact, the location of the contact point (positioning), and the characteristics of the contact, i.e., the force it is subjected to (characterization). Therefore, the strain gauge mentioned in the previous section is used in conjunction with the basic assumptions of specific stress detection, such as calculating the force between the worktable and the object. Specific methods include installing sensors on the environment, installing testing instruments on the robot's wrist, and using the transmission device as a sensor.
Proximity sensor
Due to the increased speed of robot movement and the potential damage caused by loading and unloading objects, it is necessary to know the prior information about the object's position within the robot's work area and to plan appropriate trajectories. Therefore, it is necessary to apply remote sensing methods to measure proximity. Proximity sensors are divided into passive and active sensors, so in addition to natural signal sources, artificial signal transmitters and receivers may also be required.
Ultrasonic proximity sensors are used to detect the presence of objects and measure distances. They cannot measure distances smaller than 30-50 cm, but have a relatively large range, making them suitable for use on mobile robots and in the grippers of large robots. They can also be used to create ultrasonic navigation systems.
Infrared proximity sensors are very small, only a few cubic centimeters in size, so they can be installed on robot grippers.
acoustic sensor
Sound sensors are used to sense and interpret sound waves in gases (non-contact sensing), liquids, or solids (contact sensing). The complexity of sound sensors ranges from simple detection of the presence of sound waves to complex analysis of sound frequency, and even the discrimination of individual words and phrases in continuous natural language.
Contact or non-contact temperature sensors
In recent years, thermoelectric television cameras have been widely used in robots. In addition to commonly used resistance temperature detectors (RTDs) and thermocouples, progress has also been made in measuring and sensing temperature images.
Slip sensor
Used to detect object slippage. When a robot is required to grasp an object with unknown properties, the optimal grip force must be determined. Therefore, it is necessary to detect the object slippage signal that occurs when the grip force is insufficient. Using this signal, the robot can firmly grasp the object without damaging it.
Currently, there are gliss sensors that utilize optical systems and gliss sensors that utilize crystal receivers. The latter's detection sensitivity is independent of the gliding direction.
Distance sensor
Distance sensors used in intelligent mobile robots include laser rangefinders (which can also measure angles) and sonar sensors, which have seen development in recent years.
Visual sensors
Emerging in the late 1950s, machine vision has developed rapidly and is one of the most important sensors in robotics. Starting in the 1960s, machine vision initially processed the world of building blocks, later expanding to handle the real world outdoors. Practical vision systems emerged after the 1970s. Vision generally includes three processes: image acquisition, image processing, and image understanding. Relatively speaking, image understanding technology is still quite underdeveloped.
Key specifications of the sensor:
Dynamic range refers to the range that a sensor can detect. For example, a current sensor that can measure currents from 1mA to 20A has a measurement range of 10log(20/0.001) = 43dB. If the sensor's input exceeds its measurement range, the sensor will not display the correct measurement value. For example, an ultrasonic sensor cannot measure objects at close range.
Resolution: Resolution refers to the smallest difference a sensor can measure. For example, a current sensor might have a resolution of 5mA, meaning it cannot detect current differences smaller than 5mA. Naturally, higher resolution sensors are more expensive.
Linearity: This is a very important metric for measuring the relationship between sensor input and output.
Frequency: refers to the sampling rate of the sensor. For example, an ultrasonic sensor with a sampling rate of 20Hz means it can scan 20 times per second.
Requirements and selection of robot sensors
The selection of robot sensors depends on the robot's operational needs and application characteristics. The requirements for the robot's sensory system are the basic basis for selecting sensors.
High precision and good repeatability;
Good stability and reliability
Strong anti-interference ability;
Lightweight, compact, and easy to install.
summary
All robot sensors are closely related to signal transformation and processing after signal detection. Moreover, the software workload for transformation and processing is huge, and it is also integrated with information technologies such as artificial intelligence. Therefore, this article is just a starting point and needs to be explored in depth.
