Method for selecting a sequence of signals for controlling an actuator

Abstract

The inventions relate to the field of computing and describe a method for selecting a sequence of signals for controlling an actuator, as well as embodiments of devices for implementing same and uses of said method. The method for selecting a sequence of signals for controlling an actuator comprises steps in which a control signal is generated and sent to an actuator. The state of the actuator is registered by a sensor and data from the sensor is transmitted to a computing device. If the received state of the actuator is different from a target state, the computing device calculates a new signal using a neural network for calculating new signal parameters. The result is an improvement in the quality, speed and stability with which technical systems are controlled.

Description

[0001] METHOD OF SELECTING A SEQUENCE OF SIGNAL FOR CONTROLLING AN EXECUTIVE DEVICE

[0002] AREA OF TECHNOLOGY

[0003] The present technical solution relates to the field of computer technology, in particular, to methods for selecting a sequence of control signals for controlling an actuator that does not provide feedback; in particular, one of the applications of the said method is controlling displays that operate on the basis of the principle of electrophoresis.

[0004] LEVEL OF TECHNOLOGY

[0005] A solution selected as the closest analogue is known from the prior art, US 2022138986 (A1), published 05.05.2022. This solution is intended for configuring a color measurement device. In a specific implementation, a method for matching color measurements is provided by implementing a one-step process of calibration and color conversion using an artificial neural network (ANN). As an example, raw readings from a color measurement device are converted into a specific color space, such as L*a*b, directly via an ANN. Such an artificial neural network is trained to ensure that the color measurement output data from the color measurement device matches the color measurement output data from the target color measurement device.At the same time, the said invention solves a qualitatively different problem: with the help of the ANN, uniformity of colorimetric measurements is achieved on different devices, while the proposed solution solves the problem of finding optimal parameters for forming an image on a screen operating on the phenomenon of electrophoresis (hereinafter referred to as an electronic ink display).

[0006] The proposed technical solution is aimed at eliminating the shortcomings of the current level of technology and differs from previously known ones in that the proposed solution performs a more qualitative and effective selection of the sequence of control signals for actuators that do not provide direct feedback. Due to the lack of feedback, the control device does not have data on how the control signal passed and also does not have the ability to adapt to changing external conditions.The proposed solution is aimed at creating a method for controlling devices that do not provide feedback, based on the introduction of an observing device (sensor) and a neural network into the technical system, with the help of which the control device is trained to issue control signals that will most accurately lead to the required results, for example, waveforms can be used as such a signal to display an image on a display operating on the phenomenon of electrophoresis (a display based on electronic ink).

[0007] ESSENCE OF THE INVENTION

[0008] The technical problem, which the claimed solution is aimed at solving, is the creation of a method for selecting a sequence of signals to bring the actuator to the required state, for example, to display an image on a display operating on the phenomenon of electrophoresis. At the same time, such a method can be used in any technical system without direct feedback, or in systems where direct feedback is supplemented by other sources of information about the state of the systems - for example, in unmanned aerial vehicles, robotic systems, audio systems, etc.

[0009] The technical result consists in improving the quality, speed, and stability of control of technical systems.

[0010] The claimed technical result is achieved by implementing a method for selecting a sequence of signals for changing a controlled device, comprising the following stages: a) the computing device (4), via the controller (1), generates a control signal for the transition of the actuator (2) from the current state to the expected state; b) the computing device (4), via the controller (1), sends the generated control signal to the actuator (2) for the transition of the actuator (2) from the current state to the expected state; c) the sensor (3) records the state of the actuator (2) obtained as a result of processing the control signal; d) the data on the state of the actuator (2) received from the sensor (3) are transmitted to the computing device (4);e) if the received state of the actuator (2) differs from the target one, the computing device (4) calculates a new signal, namely: i) the neural network connected to the computing device (4) receives the current and target states of the actuator (2) as input and calculates the parameters of the new signal; ii) the calculated signal is sent to the actuator (2); iii) the sensor (3) reads the received state of the actuator (2) and transmits it to the input of the computing device (4); iv) then, if there is a difference between the expected state and the received state of the actuator (4), the weights of the neurons of the neural network are updated by means of the algorithm for updating the weights of the neural network; v) after updating the weights, the neural network calculates new parameters of the signal and a new control cycle is started with sending a signal to the actuator (2).

[0011] In a particular embodiment of the described solution, the method according to item 1, in which a sensor or group of sensors reading the state of several areas of the actuator simultaneously is used as a sensor of the device or several devices.

[0012] In a particular embodiment of the described solution, the method according to paragraph 1, characterized in that simultaneous reading of at least one or more sections of the actuator (2) is performed.

[0013] In a particular embodiment of the described solution, a device for implementing the method according to item 1, in which the actuator (2) is represented by a screen (display).

[0014] In a particular embodiment of the described solution, the method according to item 1, in which the controller (1) and the sensor (3), as well as the neural network, are implemented as part of a computing device (4) and have a software and hardware design.

[0015] In a particular embodiment of the described solution, the method according to item 1 is used to search for and record waveform parameters for controlling displays (screens) operating on the phenomenon of electrophoresis.

[0016] In a particular embodiment of the described solution, a device for implementing the method according to item 1, in which the actuator (2) is represented by a screen (display), and the sensor (3) is represented by a device for recording an image. In a particular embodiment of the described solution, a device for implementing the method according to item 1, in which the actuator (2) is represented by a screen (display), and the sensors are represented by a combination of sensors of various types, including one or more machine vision devices, one or more devices recording the surface temperature of the screen (display).

[0017] In a particular embodiment of the described solution, the device according to item 4 additionally contains at least one light source for illuminating the surface of the screen (display).

[0018] In a particular embodiment of the described solution, the device according to item 4 additionally contains at least one diffusing chamber (lightbox) for stabilizing lighting conditions.

[0019] In a particular embodiment of the described solution, the device according to item 4, additionally containing at least one light sensor for calibrating illumination levels and monitoring the stability of the operating results of the device.

[0020] In a particular embodiment of the described solution, the device according to item 4 has a case, or a caseless, or a desktop, or a mobile, or a stationary design, depending on the requirements of the end user and / or the conditions of use.

[0021] In a particular embodiment of the described solution, the device according to item 4 is equipped with an integrated computing unit for autonomous operation without connection to other computing systems.

[0022] In a particular embodiment of the described solution, the device according to item 4 does not have an integrated computing unit and is controlled by an external computing device via a wired or wireless communication channel.

[0023] In a particular embodiment of the described solution, the device according to item 4, having at least one place for placing and connecting one or more screens (displays) operating on the phenomenon of electrophoresis, wherein the displays are connected to the device via various standard physical connectors such as FPC, SlimPort, MHL, FFC and others, wherein the specific method of connecting the displays depends on the standard used on the display itself, wherein the device can support several standards at once for connecting a wide range of displays of different standards.

[0024] In a particular embodiment of the described solution, a device implementing the method according to item 1, intended for searching for waveforms for displays (screens) operating on the phenomenon of electrophoresis (screen / display on electronic ink, EPD, e-ink screen / display), determining combinations of waveforms with new properties that improve the characteristics of images output using such improved waveforms.

[0025] In a particular embodiment of the described solution, the method specified in paragraph 1 is used to calibrate and improve the quality of the image of information on any displays and display means capable of displaying visual information according to specified parameters, including such as liquid crystal displays, organic diode displays, mechanical, electromechanical, photonic crystal displays, as well as other others, where, using the described sequence of actions, maximum correspondence is achieved between the displayed image and the actual image obtained on the display device.

[0026] In a particular embodiment of the described solution, the method specified in paragraph 1 is used for calibrating and improving the quality of reproduction in audio systems of any type, wherein the computing device can be represented by a digital-to-analog converter, the sensor can be represented by one or more microphones, the actuator can be represented by a sound amplifier with sound sources that adapt to the acoustic properties of rooms and surrounding objects, minimizing the negative acoustic properties of such rooms and objects, taking into account the actual sound reflections and the ability to absorb sound during digital-to-analog sound conversion to achieve maximum sound fidelity.

[0027] In a particular embodiment of the described solution, the method specified in paragraph 1 is applied in control systems for robotic devices, remotely controlled and / or unmanned technical means, which consists in the fact that a controller or control unit of the technical means is used as a computing device, a power plant (for example, propulsion systems) and executive control bodies (for example, rotary mechanisms, deflectable aerodynamic surfaces) are used as an actuator, and a gyroscope, accelerometer, lidar, satellite navigation signal receiver, radio altimeter are used as sensors, with the help of which the position, orientation and speed of the technical means are determined, and the result of applying the method is training the computing device to generate higher quality control signals for stable task execution and more accurate maneuvering.

[0028] In a particular embodiment of the described solution, the method specified in paragraph 1 is applied in intelligent heating, ventilation, air conditioning (HVAC) systems, wherein the computing device can generate control signals for air conditioning, heating or ventilation systems, and fans, dampers or valves that regulate air flow and temperature are used as actuators, and, for example, temperature sensors and humidity sensors are used as sensors, which are used to monitor environmental conditions, and the result of applying the method is training the computing device to generate higher quality control signals to improve energy efficiency and maintain optimal heating, ventilation or air conditioning conditions.

[0029] In a particular embodiment of the described solution, the method specified in paragraph 1 is applied in intelligent motorized prosthetic limbs, wherein the computing device can generate control signals for the movement of the prosthetic elements, for example, motors or other mechanisms controlling the movement of the prosthesis are used as actuators, muscle sensors, force sensors or encoders used to collect data on user commands and on the states of the moving elements of the prosthesis and the limb as a whole can be used as sensors, and the result of applying the method is training the computing device to generate higher quality control signals for more natural and intuitive control of the prosthesis.

[0030] In a particular embodiment of the described solution, the method specified in paragraph 1 is applied in robotic systems, wherein the computing device can generate control signals for actuating robotic arms, grippers or legs, and motors or servo drives that set the robot components in motion can be used as actuators, encoders, cameras or force sensors can be used as sensors, with the help of which the productivity and positions of the components of the robotic system are measured, and the result of applying the method is training the computing device to generate higher-quality control signals for controlling the motor to increase accuracy, speed or energy efficiency. In a particular embodiment of the described solution, the method specified in paragraph 1 is applied.1 method in autonomous vehicles, wherein the computing device can generate control signals for steering, acceleration and braking, electric motors, hydraulic systems or other mechanisms controlling the movement of the vehicle can be used as actuators, cameras, gyroscopes, accelerometers, LIDAR and GPS can be used as sensors, which are used to collect data on the position of the vehicle and its environment, and the result of using the method is training the computing device to generate higher quality control signals for safe and efficient driving in various conditions.

[0031] DESCRIPTION OF DRAWINGS

[0032] The implementation of the invention will be described further in accordance with the attached drawings, which are presented to explain the essence of the invention and in no way limit the scope of the invention. The following drawings are attached to the application:

[0033] Fig. 1 illustrates an example of the implementation of a method for selecting a signal sequence for displaying an image on a display based on electronic ink.

[0034] DETAILED DESCRIPTION OF THE INVENTION

[0035] In the following detailed description of the embodiment of the invention, numerous implementation details are set forth in order to provide a clear understanding of the present invention. However, it will be apparent to one skilled in the art how the present invention may be used with or without these implementation details. In other instances, well-known methods, procedures, and components have not been described in detail in order not to unnecessarily obscure the features of the present invention.

[0036] In addition, it will be clear from the above presentation that the invention is not limited to the given implementation. Numerous possible modifications, changes, variations and replacements, preserving the essence and form of the present invention, will be obvious to specialists qualified in the subject area. The present technical solution, due to its implementation, ensures high-quality and effective selection of a signal sequence (waveform) for displaying an image on a display based on electronic ink).

[0037] An e-ink display consists of a thin layer of "cones" containing black and white particles. The black and white particles have an electrical charge (for example, black particles are positively charged and white particles are negatively charged).

[0038] Under the layer of “cones” there is a substrate that is capable of transmitting a positive or negative charge.

[0039] Thus, if, for example, the substrate is charged with a positive charge, then white particles will be attracted to it, and black particles will be repelled. And, therefore, for the viewer the color of the display will become black.

[0040] Thus, by sending a certain charge to each pixel (capsule), it is possible to obtain stable changes in the color of such a pixel over time, since statically charged particles stably maintain their position over a long period of time, which is the main feature of all EPD displays - the energy independence of the graphics displayed on the display.

[0041] To obtain shades of gray, it is necessary to transfer the charge in such a way that the black and white particles are mixed in a certain proportion. This is achieved by transmitting a complex sequence of electrical impulses to the capsule electrode. Such a sequence of impulses is called a waveform.

[0042] For a 16-color display (a display with 16 shades of gray), a specific waveform is selected to transition from each color to each.

[0043] For example, to go from color 4 to color 9 - one waveform;

[0044] For the transition from color 11 to color 2, another waveform is used. Thus, all color transitions can be considered as combinations of non-transitive pairs of transitions from color to color within a set of 16 colors.

[0045] In total there are (16! / (2! * (16-2)!)) x 2 = 240 such waveforms.

[0046] Moreover, the inertness of white and black particles depends on the temperature of the display, which depends on many factors, mainly the ambient temperature. Temperature also significantly affects the viscosity of the liquid in which the particles float.

[0047] Thus, for the correct operation of the display, it is necessary that for each small change in the display temperature, there is its own set of 256 waveforms, which will take into account, at least, these factors as reliably as possible. As a rule, temperatures from 10 to 30 degrees are supported, with a step of 2 degrees. Therefore, for the correct operation of the display, based on electronic ink, it is necessary to select about (30-10) / 2 * 240 = 2400 waveforms.

[0048] An additional factor complicating display control is the mutual influence of the static charge of the particles in the capsule on the particles in neighboring capsules, when the charge imparted to the particles in one capsule causes the movement of the particles in the neighboring capsule.

[0049] It should also be noted that due to errors in the physical processes involved in display production, physical characteristics will always vary slightly from batch to batch.

[0050] Thus, for each new batch, specific “fine-tuning” of the waveform is required for a particular batch.

[0051] Selecting such waveforms is a complex task that requires a lot of resources.

[0052] To simplify the task of selecting waveforms, the implementation of this technical solution is proposed.

[0053] Fig. 1 illustrates a method for selecting a signal sequence for displaying an image on an electronic ink display, where:

[0054] 1 - Screen controller, based on electronic ink;

[0055] 2 - Display based on electronic ink;

[0056] 3 - Sensor (a device for accurately determining the color displayed on the screen, such as a colorimeter or cameras of various designs);

[0057] 4 - Computing device (as a special case: computers with central processors of various architectures, such as x86, RISC, RISC-V and others).

[0058] An example of the implementation of this technical solution.

[0059] The computing device (4), via the controller (1), sends to the electronic ink display (2) the generated waveform for the transition from the current color to the target color.

[0060] Then, using the sensor (3), it reads the obtained result.

[0061] After this, if the obtained result differs from the target, the computing device applies machine learning methods to select a new waveform.

[0062] The following can be used as a sensor:

[0063] Colorimeter as a device for measuring light intensity;

[0064] A camera that can monitor one or many areas of the screen at the same time;

[0065] Any device that measures light intensity. Light intensity measurement can be coupled with ghosting recognition algorithms. In this case, the accuracy of color detection can be improved by knowing the color difference between adjacent areas. An example of such an algorithm was described in the article:

[0066] [https: / / www.researchgate.net / publication / 346358764_A_Convolutional_Neural_Ne twork_for_Ghost_lmage_Recognition_and_Waveform_Design_of_Electrophoretic_Display s]

[0067] Unlike prior-art approaches, machine learning methods are able to select not only the waveform itself, but also other parameters, such as quantity, frequency, amplitude, applied voltage, and the polarity of the pulses that make up the waveform in their relationship with each other, as well as search for new previously unknown hidden relationships between changes in some parameters and their mutual influence on others. This gives an advantage in the potential quality and speed of image display.

[0068] To speed up the algorithm, it is possible to apply different waveforms to small areas of the screen. Thus, during the operation of the algorithms, steps 1-3 can work simultaneously, independently of each other, parallelizing the algorithm. Also, for different areas of the screen, different algorithms can be applied in parallel, working according to the general description 1-3.

[0069] Machine learning methods may include, but are not limited to, the following:

[0070] 1. Classical fully connected neural network

[0071] Architecture

[0072] Neural network input (Input Layer) - 3 neurons:

[0073] Current color (obtained from the colorimeter);

[0074] The target color is the one that should be obtained after applying the generated waveform;

[0075] Panel temperature (obtained from the temperature sensor);

[0076] Hidden layers - n layers with width from k to I

[0077] The output layer of the neural network is j neurons, where j is the length of the target waveform.

[0078] Each output neuron represents an amplitude. Thus, a sequence of j neurons represents a waveform.

[0079] The following functions can be used as the activation function of the network neurons (but are not limited to):

[0080] Sigmoid; Hyperbolic tangent;

[0081] ReLU.

[0082] The neural network is trained as follows:

[0083] The neural network receives current data and a target color as input, using which it generates a waveform;

[0084] The generated waveform is applied to an e-ink display.

[0085] The sensor reads the received color. Then, if there is a difference between the target color and the received color, the weights of the neurons of the neural network are updated using the algorithm for updating the weights of the neural network.

[0086] Also, one of the neural network regularization methods can be used during training.

[0087] • The neural network update algorithm can be one of the following (but is not limited to):

[0088] • Backpropagation (http: / / www.cs.utoronto.ca / ~hinton / absps / naturebp.pdf)

[0089] • The HSIC Bottleneck (Hilbert-Schmidt Independence Criterion)

[0090] • Evolutionary Algorithm (eg genetic algorithm)

[0091] • Online Alternating Minimization with Auxiliary Variables

[0092] • Decoupled Neural Interfaces Using Synthetic Gradients Neural Network Regularization Algorithms

[0093] The following algorithm can be chosen as a regularization algorithm for the neural network:

[0094] • L1 Regularization;

[0095] • L2 Regularization;

[0096] • Dropout.

[0097] 2. Recurrent neural network A recurrent neural network generates a sequence of waveform elements, receiving the previous element as input at each iteration and generating the next one.

[0098] Neural network input (Input Layer) - 4 neurons:

[0099] • Current color (obtained from the colorimeter);

[0100] • Target color - the one that should be obtained after applying the generated waveform;

[0101] • Panel temperature (obtained from the temperature sensor);

[0102] • The previous value of the sequence element.

[0103] Hidden layers - a sequence of: n layers with widths from k to I; a recurrent module (or several recurrent modules located on one layer);

[0104] The output layer of the neural network is 1 neuron, which is a generated element of the waveform. This element is fed to the input of the recurrent neural network to generate the next element.

[0105] A recurrent module can be one of the following modules:

[0106] • Classical recurrent module;

[0107] • LSTM (long short-term memory) module;

[0108] • GRU (Gated Recurrent Unit) module.

[0109] The neural network is trained as follows:

[0110] The length of the waveform (the target number of elements to be generated) is determined.

[0111] In a loop, the number of iterations of which is equal to the target length of the waveform, the neural network receives the current data and the target color as input, as well as the value of the waveform element from the previous loop iteration (at the zero iteration it receives a predetermined constant, for example 0). Thus, at each iteration of the loop, the next waveform element is generated. By the time the loop finishes working, we receive the generated waveform.

[0112] The generated waveform is applied to an e-ink display.

[0113] The sensor reads the received color. Then, if there is a difference between the target color and the received color, then the weights of the neurons of the neural network are updated using the neural network weight update algorithm.

[0114] Also, one of the neural network regularization methods can be used during training.

[0115] The neural network update algorithm can be one of the following (but is not limited to):

[0116] Backpropagation;

[0117] The HSIC Bottleneck (Hilbert-Schmidt Independence Criterion);

[0118] Evolutionary Algorithm (eg genetic algorithm);

[0119] Online Alternating Minimization with Auxiliary Variables;

[0120] Decoupled Neural Interfaces Using Synthetic Gradients.

[0121] Neural Network Regularization Algorithms

[0122] The following algorithm can be chosen as a regularization algorithm for the neural network:

[0123] L1 Regularization; L2 Regularization;

[0124] Dropout.

[0125] Genetic algorithm. With the help of a genetic algorithm, a multitude of genotypes are derived, which compete with each other through the mechanism of crossing, mutation and natural selection. The genotype is a vector, which is a waveform. Thus, after working through many generations, a generation of genotypes with maximum efficiency is “derived”.

[0126] Operating algorithm:

[0127] 1. A population of N genotypes of length M filled with random numbers is generated. The genotype is the waveform vector.

[0128] 2. Waveforms are applied to the screen (for parallelism, each waveform can be applied to a small area of ​​the screen).

[0129] 3. The colorimeter reads the values ​​from the obtained areas.

[0130] 4. The fitness function is calculated, which is the difference between the target and the resulting color.

[0131] 5. A selection of n genotypes with the highest fitness function is carried out from the population.

[0132] 6. Among them, crosses with a crossing coefficient k and mutations with a mutation rate gl are produced. In this way, a new population of N genotypes is generated.

[0133] Steps 1-6 are repeated P times.

[0134] After which the most adapted genotype represents the most optimal waveform.

[0135] In particular, the described solution can be used in other applied areas of technology:

[0136] 1. To create intelligent audio calibration systems for consumer and commercial applications that will adapt to different environmental conditions to improve sound quality and maintain its stability in various situations.

[0137] 2. For integration with robotic, cybernetic, biomechanical manipulators and limb prostheses, in which, by adjusting the operation of servo drives and encoders, an increase in useful characteristics will be achieved, such as efficiency, energy efficiency, strength, number of working cycles before wear, etc.

[0138] The present application materials present a preferred disclosure of the implementation of the claimed technical solution, which should not be used as limiting other, particular embodiments of its implementation that do not go beyond the scope of the requested scope of legal protection and are obvious to specialists in the relevant field of technology.

Claims

FORMULA 1. A method for selecting a sequence of signals for controlling an actuator, comprising the following stages: a) a computing device (4), via a controller (1), generates a control signal for the actuator (2) to transition from the current state to the expected state; b) a computing device (4), via a controller (1), sends the generated control signal to the actuator (2) for the actuator (2) to transition from the current state to the expected state; c) a sensor (3) records the state of the actuator (2) obtained as a result of processing the control signal; d) data on the state of the actuator (2) received from the sensor (3) are transmitted to the computing device (4);e) if the received state of the actuator (2) differs from the target, the computing device (4) calculates a new signal, namely: i) the neural network connected to the computing device (4) receives the current and target state of the actuator (2) as input and calculates the parameters of the new signal; ii) the calculated signal is sent to the actuator (2); Hi) the sensor (3) reads the received state of the actuator (2) and transmits it to the input of the computing device (4); iv) then, if there is a difference between the expected state and the received state of the actuator (4), the weights of the neurons of the neural network are updated by means of the algorithm for updating the weights of the neural network; v) after updating the weights, the neural network calculates new signal parameters and a new control cycle is started with sending a signal to the actuator (2).

2. The method according to claim 1, in which a sensor or group of sensors that read the state of several areas of the actuator simultaneously is used as a sensor of the device or several devices.

3. The method according to claim 1, characterized in that simultaneous reading of at least one or more sections of the actuator (2) is performed.

4. A device for implementing the method according to item 1, in which the actuator (2) is represented by a screen (display).

5. The method according to item 1, in which the controller (1) and the sensor (3), as well as the neural network, are implemented as part of a computing device (4) and have a software and hardware design.

6. The method according to item 1, used for searching and recording waveform parameters for controlling displays (screens) operating on the electrophoresis phenomenon.

7. A device for implementing the method according to item 1, in which the actuator (2) is represented by a screen (display), and the sensor (3) is represented by a device for recording an image.

8. A device for implementing the method according to item 1, in which the actuator (2) is represented by a screen (display), and the sensors are represented by a combination of sensors of various types, including one or more machine vision devices, one or more devices recording the temperature of the surface of the screen (display).

9. The device according to claim 4, further comprising at least one light source for illuminating the surface of the screen (display).

10. The device according to item 4, additionally containing at least one diffusing chamber (light box) for stabilizing lighting conditions.

11. The device according to claim 4, further comprising at least one light sensor for calibrating light levels and monitoring the stability of the operating results of the device.

12. The device according to item 4, having a case, or a caseless, or a desktop, or a mobile, or a stationary design depending on the requirements of the end user and / or the conditions of use.

13. The device according to item 4, equipped with an integrated computing unit for autonomous operation without connection to other computing systems.

14. The device according to item 4, which does not have an integrated computing unit and is controlled by an external computing device via a wired or wireless communication channel.

15. The device according to item 4, having at least one place for placing and connecting one or more screens (displays) operating on the phenomenon of electrophoresis, wherein the displays are connected to the device via various standard physical connectors such as FPC, SlimPort, MHL, FFC and others, wherein the specific method of connecting the displays depends on the standard used on the display itself, wherein the device can support several standards at once for connecting a wide range of displays of different standards.

16. A device implementing the method according to item 1, intended for searching for waveforms for displays (screens) operating on the phenomenon of electrophoresis (electronic ink screen / display, EPD, e-ink screen / display), determining combinations of waveforms with new properties that improve the characteristics of images output using such improved waveforms.

17. The use of the method specified in paragraph 1 for calibrating and improving the quality of the image of information on any displays and display devices capable of displaying visual information according to specified parameters, including such as liquid crystal displays, organic diode displays, mechanical, electromechanical, photonic crystal displays, as well as other devices, where, using the described sequence of actions, maximum correspondence is achieved between the displayed image and the actual image obtained on the display device.

18. The use of the method specified in paragraph 1 for calibration and improving the quality of reproduction in audio systems of any type, wherein the computing device can be represented by a digital-to-analog converter, the sensor can be represented by one or more microphones, the actuator can be represented by a sound amplifier with sound sources that adapt to the acoustic properties of rooms and surrounding objects, minimizing the negative acoustic properties of such rooms and objects, taking into account the actual sound reflections and sound absorption capability during digital-to-analog sound conversion to achieve maximum sound fidelity.

19. The use of the method specified in paragraph 1 in control systems for robotic devices, remotely controlled and / or unmanned technical means, consisting in the fact that a controller or control unit of the technical means is used as a computing device, a power plant (for example, propulsion systems) and executive control bodies (for example, rotary mechanisms, deflectable aerodynamic surfaces) are used as an actuator, and a gyroscope, accelerometer, lidar, satellite navigation signal receiver, radio altimeter are used as sensors, with the help of which the position, orientation and speed of the technical means are determined, and the result of using the method is training the computing device to generate higher quality control signals for stable task execution and more accurate maneuvering.

20. The use of the method specified in paragraph 1 in intelligent heating, ventilation, air conditioning (HVAC) systems, wherein the computing device can generate control signals for air conditioning, heating or ventilation systems, and fans, dampers or valves that regulate air flow and temperature are used as actuators, and temperature sensors and humidity sensors are used as sensors, for example, which are used to monitor environmental conditions, and the result of using the method is training the computing device to generate higher quality control signals to improve energy efficiency and maintain optimal heating, ventilation or air conditioning conditions.

21. The use of the method specified in paragraph 1 in intelligent motorized prosthetic limbs, wherein the computing device can generate control signals for the movement of the prosthetic elements, for example, motors or other mechanisms that control the movement of the prosthesis are used as actuators, muscle sensors, force sensors or encoders used to collect data on user commands and on the states of the moving elements of the prosthesis and the limb as a whole can be used as sensors, and the result of using the method is training the computing device to generate higher quality control signals for more natural and intuitive control of prostheses.

22. The use of the method specified in paragraph 1 in robotic systems, wherein the computing device can generate control signals for actuating robotic arms, grippers or legs, and motors or servo drives that actuate the robot components can be used as actuators, encoders, cameras or force sensors can be used as sensors, with the help of which the performance and positions of the components of the robotic system are measured, and the result of using the method is training the computing device to generate higher quality control signals for controlling the motor to increase accuracy, speed or energy efficiency.

23. The use of the method specified in paragraph 1 in autonomous vehicles, wherein the computing device can generate control signals for steering, acceleration and braking, electric motors, hydraulic systems or other mechanisms controlling the movement of the vehicle can be used as actuators, cameras, gyroscopes, accelerometers, LIDAR and GPS can be used as sensors, which are used to collect data on the position of the vehicle and its environment, and the result of using the method is training the computing device to generate higher quality control signals for safe and efficient driving in various conditions.

Images