Different types of sensors in robotics are used for different purposes. Sensors are needed for robots to receive information about themselves and their physical environment. Now there are a great many: from sensors of mechanical quantities (linear, angular displacements, distance, acceleration, forces and moments) to technical vision systems, temperature, current, and voltage meters, light intensity, radioactive and magnetic fields, acoustic sensors, water detectors, and gas analyzers and others. Moreover, they all work on different physical principles that determine the range of conditions in which the required quality of measurements can be ensured.
For example, the distance to surrounding objects can be measured using sonars, infrared range finders, ToF cameras, radars or lidars (scanning laser range finders), that is, using radiations of different wavelengths. And if the former go blind on fuzzy surfaces, the latter begins to lie with background IR radiation, and still, others when measuring the distance to dark surfaces, then the lidars compare favorably in resolution and range, but they clearly lose in price. It is also clear that the choice of sensors depends on the environment of their intended use.
The reason is clear: accuracy and speed are needed, especially when trying to assess a dynamically changing environment (for example, distances to moving obstacles in autonomous navigation). And finding a middle ground between price and quality is a non-trivial scientific and technical challenge that free researchers, startups, and industry giants are trying to solve. One big jump has already been. Serious progress has been made with the advent of microelectromechanical (MEMS) systems, and this is attributed to the mass production of smartphones. If earlier robots used expensive and overall accelerometers and gyroscopes, then using MEMS sensors now even consumer robotics can be equipped with sensors. This is an example of enabling technology and how one area of technology affects the development of another.
Different types of sensors in robotics
In a sense, robots try to copy the senses that humans or animals have. But often robots have much more advanced systems. So, the vestibular apparatus of a person fixes a change in the position of the body or head, but there is no organ that would suggest how many angular minutes the knee or elbow bent or at what distance the object is from us with micron accuracy. Man does not need this, and in the current paradigm of development of robotics, this information does not interfere. Other examples may be related to magnetic field or radiation detectors.
In general, robots can be divided into two types: locomotion and manipulation. The main task of the former is to move by themselves and move the payload or person over considerable distances, like drones, unmanned vehicles or boats do. The main task of the sensors, in this case, is to determine the robot’s own position in space, for example, based on data from odometry or global positioning systems, as well as location relative to surrounding objects. You can also add linear and angular acceleration sensors that provide a sense of balance, that is, orientation in a gravitational field.
The task of manipulation robots, which should functionally imitate hands, is to perform various operations with objects. Here the kinesthetic sensation that gives proprioceptive information, that is, feelings of position, movement, and strength, comes to the fore. That is, we need different types of sensors in robotics that allow us to determine the current configuration and speed of individual parts of the robot. As well as we need tactile and force-moment sensors. The latter are especially in demand to ensure reliable capture of objects of manipulation, as well as control of the forces of interaction with objects, environment, and man, for example, to perform contact operation efficiently and not damage the robot or injure a person who is nearby or in direct contact.
Of course, for all of these robots can be involved a huge number of auxiliary, “service” sensors from those listed earlier, the use of which depends on the particular application. Some of them give information about the internal state of the system (interoception), and some – about the environment (exteroception). In this context, it is worth mentioning separately that sensors for organizing human-robot interaction are extremely important.
Trends in robotics sensory
One of the current trends in the development of sensors for robots is the force moment sensation, the development of tactile sensors. Modern handling robots are less and less placed in separate cells behind iron fences or infrared curtains, not to mention service or personal robots. Progress is moving towards the creation of robots that can effectively and safely work in a dynamic, unstructured environment when it is impossible to arrange everything strictly in places once and for all and in close contact with humans.
Today, a lot of money has invested in making sensors of force-moment sensing affordable. Today, such strain-strain membranes, which can be localized in the joints of robots or installed between a robot and a working tool, cost tens of thousands of dollars, which significantly hinders their widespread adoption. Among manufacturers of such sensors now can be noted ATI, FUTEK, Kistler, and Hitec.
Attempts to replace such expensive elements are being made even now for a limited range of applications, as KUKA does, for example, using the two encoder schemes. New types of gauges are also proposed. For example, distributed sensors are being developed and tested, which are called artificial skin. The direction of haptic – feedback control by force – is powerfully developing in the USA, Switzerland, Germany, Korea, China, Japan, where there are large laboratories working in this field.