An Adaptive Image Sampling Method for Visual Perception Control Systems

Abstract

This invention discloses an adaptive image sampling method for visual perception control systems, relating to the field of visual perception technology. It jointly designs a feedback control strategy and a visual information sampling strategy with the goal of ensuring control system stability. The method determines which visual perception strategy to employ by characterizing the relationship between visual sampling resolution, image processing delay, and state estimation error. This invention allows for flexible adjustment of the visual perception strategy of the control system, effectively expanding the scope for improving control performance and enabling both fast and accurate control. It ensures stable operation of the control system, avoiding property damage and wasted manpower due to equipment failure caused by instability.

Description

Technical Field

[0001] This invention relates to the field of visual perception technology, and in particular to an image adaptive sampling method for visual perception control systems. Background Technology

[0002] Visual perception technology captures images and videos of industrial sites using devices such as cameras, and then uses image processing techniques to obtain the necessary environmental and status information. On the one hand, visual data contains extremely rich information about the environment, status, and appearance; on the other hand, it is a non-contact measurement method, applicable to high-temperature and high-pressure environments, high-speed moving objects, and other scenarios where it is inconvenient to install contact sensors. Therefore, visual perception technology is widely used in control systems for industrial manufacturing, intelligent robots, and drones. In the control loop of a control system, the camera transmits the acquired visual data to computing devices via a network for image processing and control decisions. However, due to the large volume of visual information and the complexity of processing algorithms, its data transmission and processing take much longer than traditional direct sensing methods. For control systems with high real-time requirements, this long delay caused by the visual perception process can prevent the controller from obtaining system status information in a timely manner, thereby reducing system control performance and potentially leading to control system instability. Therefore, how to reduce the negative impact of visual perception delay is of significant research importance for improving control system performance.

[0003] Current research has explored ways to mitigate the impact of visual perception latency on system performance through control algorithm design. Predictive control algorithms can predict the system's state using known system models when the controller cannot obtain real-time system state information in a timely manner. Robust control algorithms can analyze worst-case system control performance to ensure that the negative impacts of visual perception-induced latency and image processing errors are kept within a certain range. However, simple control algorithm design can only passively compensate for latency. If the visual perception strategy is inappropriate, this passive compensation method has significant limitations. For example, if the camera sampling resolution is too high, it will not only fail to significantly improve image processing accuracy but will also cause a surge in data transmission and processing latency, preventing the system from obtaining state information in a timely manner, greatly affecting system stability, and resulting in excessive waste of computing and network resources. Conversely, if the sampling resolution is too low, it may sacrifice image processing accuracy or even make it difficult to obtain the necessary state information through image processing algorithms, leading to inaccurate control. Therefore, how to design visual information sampling strategies to optimize system control performance is a crucial problem that urgently needs to be solved.

[0004] A review of existing literature revealed the most similar implementation scheme as follows: Chinese patent application number 201810795014.5, entitled "A Visual Servo Tracking Predictive Control Method for Mobile Robots," which specifically models the problem as a tracking error model with uncertain parameters, and provides prediction indices and controller solution methods based on this model. However, this scheme does not consider the impact of transmission and processing delays in the visual perception process on system stability, making it unsuitable for situations with high real-time requirements. Another similar scheme is patent application number 202010327512.4, entitled "A Distributed Visual Servo Feedback Control System Based on Embedded Processing," which specifically uses a high-speed hardware serial interface or Ethernet to directly connect each subsystem, and processes images in parallel through multiple CPU cores. However, this approach only improves the system's real-time performance by addressing aspects such as distributed system architecture, communication methods, and parallel computing, without improving the control algorithm and perception sampling strategy. Patent application number 201910093128.X, titled "Optimal PI Parameter Optimization Method and System for Time-Delay Visual Servo System," describes a method that pre-determines the system time delay through experiments and optimizes PI control parameters based on the differential quadrature method to improve system response performance. However, this approach does not optimize the system's visual perception strategy and can only passively compensate for the delay in the perception process. Patent application number 201410076993.0, titled "Adaptive Rate Control Method for Wireless Video Stream Services Based on QoE," describes a method that establishes a video user experience quality index and optimizes the video bitrate based on this index. However, the user experience quality index established by this approach is only suitable for less demanding scenarios such as video surveillance and cannot accurately reflect control performance, making it unsuitable for direct application in the design of visual perception strategies in control systems.

[0005] Therefore, those skilled in the art are dedicated to developing an image adaptive sampling method for visual perception control systems, designing visual information sampling strategies to optimize system control performance; this method allows for flexible adjustment of the visual perception strategy of the control system, effectively expanding the scope for improving control performance, enabling the system to achieve both fast and accurate control; and ensuring the stable operation of the control system, avoiding property losses and waste of manpower due to equipment failure caused by instability. Summary of the Invention

[0006] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is how to design a visual information sampling strategy to optimize system control performance.

[0007] To achieve the above objectives, the present invention provides an image adaptive sampling method for visual perception control systems, comprising the following steps:

[0008] Step 1: Model the relationship between processing error, processing delay and image resolution in visual perception tasks;

[0009] Step 2: Determine the control feedback matrix K and sampling trigger parameter matrix Φ that satisfy the system stability conditions;

[0010] Step 3: Start the system online operation; the initial time is t=0; initialize the trigger time list and the corresponding resolution setting list; set the camera to sample images at the minimum resolution at the initial time t0=0, i.e., r(t0)=r min ;

[0011] Step 4, at t=t k At time t, jump to step 5, where t k This is the k-th trigger moment; otherwise, skip to step 6.

[0012] Step 5: The camera uses a resolution r(t) k The system samples the data and transmits the images to the image processing unit via a network; the image processing unit then processes the images to obtain an estimate of the control system's state. The data is sent to the local controller; the time delay from image sampling to the local controller receiving the state estimate is τ. k ;

[0013] Step 6, at t=t k +τ k At that moment, the local controller received the state estimate. Skip to step 7; otherwise, skip to step 9.

[0014] Step 7: The feedback control module updates the state prediction.

[0015] Step 8: The sampling decision module updates the decisions regarding the sampling time and resolution configuration;

[0016] Step 9: Update control commands using the feedback control module. The actuator executes control instructions; if t <T tmnl Then skip to step 4; otherwise, if t = T tmnl Then the operation will terminate, where T tmnl The pre-set termination time.

[0017] Furthermore, in step 1, experiments are conducted to determine how the resolution of visual data affects image processing latency and state estimation error.

[0018] Further, step 1 includes the following steps:

[0019] Step 1.1: Determine the camera settings and select the image processing algorithm. This allows the system to obtain the system status values ​​required for system control by processing the images obtained from the camera.

[0020] Step 1.2: Set the camera resolution to a fixed value r and capture images. At the same time as the camera sampling, use a high-precision sensor to collect the target state variable value x. Use a unified timer to control the camera and the high-precision sensor.

[0021] Step 1.3: Obtain the estimated state value of the control system corresponding to each image sample using the image processing algorithm selected in Step 1.1. And record the time τ required for the image processing process. p The system state estimate obtained from image processing The image processing error ω is obtained by subtracting the corresponding state variable value x.

[0022] Step 1.4: Change the camera resolution r, and repeat steps 1.2 and 1.3 until all available camera resolutions have been traversed; obtain the functional relationship between all available resolutions and the upper bound of the modulus of the image processing error. The functional relationship between τ and image processing time p (r), where

[0023] Furthermore, step 1.1, the camera settings include camera model, placement location, parameter settings, and ambient light settings.

[0024] Furthermore, an alternative to step 1.2 is to not install a high-precision sensor and use the state estimation result corresponding to the highest resolution image as the state variable value x.

[0025] Furthermore, in step 2, the stability condition is given by the Lyapunov method.

[0026] Furthermore, in step 7, The calculation method is as follows: Where A is the system state matrix and B is the system input matrix.

[0027] Furthermore, in step 8, a stability boundary distance function is defined based on the system stability criterion, which is used as the objective function to dynamically adjust the visual information sampling strategy.

[0028] Furthermore, step 8 includes the following steps:

[0029] Step 8.1, let s = t k +τ k ;

[0030] Step 8.2: Determine whether the system stability condition is met at time s. in It is the quadratic error between the current state and the state at the last sampling time (x(t) - x(t)). k )) T Φ(x(t)-x(t k The upper bound of )) It is the quadratic state μx at the last sampling time. T (t k )Φx(t k The lower bound of ) is μ, which is a pre-selected weight parameter, 0 < μ < 1; if the system stability condition is not met, jump to step 8.3; otherwise, jump to step 8.4;

[0031] Step 8.3: For each image resolution r, calculate the corresponding total image transmission and processing delay. make like Then let in h The minimum sampling time interval allowed by the camera; if Then let Calculate the stability boundary distance function C(r,t) corresponding to resolution r. k ); Select C(r,t) k Maximum resolution r * and its corresponding trigger time t k+1|k , will t k+1|k t k+1|k +1、……、 Add it to the trigger time list and set the corresponding image resolution to r. * Skip to step 9;

[0032] Step 8.4: If s or a previous time is in the trigger time list, remove it and remove the corresponding resolution setting from the resolution setting list; let s = s + 1; if If so, proceed to step 8.2; otherwise, proceed to step 8.5. This is the maximum end-to-end delay;

[0033] Step 8.5, if t k +1、t k +2、……、 If none of them are in the trigger time list, then... Add it to the trigger time list and set the corresponding resolution to the minimum resolution.

[0034] Furthermore, step 8.3, The calculation method is as follows: Where τ u =αr 2 / b represents the uplink image transmission delay, τ dFor downlink state transmission delay estimation, h is the sampling interval of the controlled object, b is the network bandwidth, and αr 2 This represents the size of the image data, where α is a constant.

[0035] In a preferred embodiment of the present invention, considering that existing visual perception delay control system design methods only passively compensate for the negative impact of delay on control performance by designing control algorithms, without any design of the visual perception strategy, the improvement of control performance is greatly limited. For example, if the camera sampling resolution is too high, it will not only fail to significantly improve image processing accuracy, but will also cause a surge in data transmission and processing delays, making it impossible for the system to obtain state information in a timely manner, greatly affecting system stability; while if the sampling resolution is too low, it may sacrifice image processing accuracy, or even make it difficult to obtain the required state information through image processing algorithms, resulting in inaccurate control. The present invention aims to ensure the stability of the control system by jointly designing a feedback control strategy and a visual information sampling strategy; it determines which visual perception strategy to adopt by characterizing the relationship between visual sampling resolution, image processing delay, and state estimation error. Experiments are conducted to obtain how the resolution of visual data affects image processing delay and state estimation error; based on robust control ideas and the Lyapunov stability criterion, system stability conditions are constructed by comprehensively considering image processing, transmission delay, and state estimation error, thereby giving the range of control strategies and visual sampling strategies that can guarantee system stability in the worst case.

[0036] Existing adaptive control methods for video stream parameters mostly adjust resolution and bitrate by defining subjective indicators such as user experience quality. These methods cannot be directly transferred to the design of visual sampling strategies in visual perception control systems because their objective functions are too subjective and do not consider the impact of visual perception strategies on system control performance from a mechanistic perspective. Therefore, the actual effect will deviate from the expected system control objective, and they can only be used in scenarios with less stringent requirements, such as video surveillance. This invention characterizes the impact of visual perception strategies on system control performance from the perspective of system control mechanisms. Based on the system stability criterion, a stability boundary distance function is defined, which is used as the objective function to dynamically adjust the visual information sampling strategy. A predictable stability boundary distance function is constructed based on the stability criterion, and the optimal resolution configuration is dynamically selected using this as the optimization objective, so that the selected visual perception strategy can optimize system stability.

[0037] Most existing visual perception-based control system design methods assume optimal visual perception accuracy or employ the most accurate visual perception strategy. However, high-precision results require rich visual information and complex image processing algorithms, consuming significant network and computational resources. In reality, since control algorithms can compensate for perception errors to some extent, blindly pursuing the highest accuracy is usually unnecessary and wastes network and computational resources. This invention employs a self-triggered visual information sampling strategy, sampling visual information only when a potential violation of the system stability criterion is predicted. The system stability criterion is transformed into a locally predictable form; visual information is sampled only when the local sampling decision unit predicts that the criterion will be violated at a certain moment. This minimizes the number of visual information samples while ensuring system stability, saving network and computational resources.

[0038] Compared with the prior art, the present invention has the following obvious substantive features and significant advantages:

[0039] 1. This invention allows for flexible adjustment of the visual perception strategy of the control system, effectively expanding the scope for improving control performance and enabling the system to achieve both fast and accurate control; it ensures the stable operation of the control system and avoids property losses and waste of manpower due to equipment failure caused by instability.

[0040] 2. This invention aims to achieve a reasonable trade-off between visual perception accuracy and latency by focusing on system control performance. It can adapt to changes in system state to select the most efficient visual perception strategy, thereby optimizing system control performance and achieving system control objectives more efficiently.

[0041] 3. This invention minimizes the network and computing resources required for the transmission and processing of visual information while ensuring system stability, thus reducing resource waste; and can efficiently complete more data transmission and processing tasks when network and computing resources are limited.

[0042] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0043] Figure 1 This is a block diagram of a visual perception control system according to a preferred embodiment of the present invention;

[0044] Figure 2 This is an offline design flowchart of a preferred embodiment of the present invention;

[0045] Figure 3 This is a flowchart illustrating the online operation of a system according to a preferred embodiment of the present invention;

[0046] Figure 4 This is a sampling decision flowchart of a preferred embodiment of the present invention. Detailed Implementation

[0047] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0048] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of some components has been appropriately exaggerated in the drawings.

[0049] The block diagram of the vision-based control system in this embodiment is as follows: Figure 1 As shown, the offline design flowchart is as follows: Figure 2 As shown, the system online operation flowchart is as follows: Figure 3 As shown, the sampling decision process of the system running online is as follows: Figure 4 As shown.

[0050] Step 1: Model the relationship between processing error, processing latency, and image resolution in visual perception tasks. For example... Figure 2 As shown, it includes the following steps:

[0051] Step 1.1: Determine the camera model, placement, parameter settings, and ambient light settings. Select an image processing algorithm that can process the images obtained by the camera to obtain the system state values ​​required for system control.

[0052] Step 1.2: Set the camera resolution to a fixed value r and capture images. At the same time as the camera sampling, use a high-precision sensor to collect the target state variable value x. Use a unified timer to control the camera and the high-precision sensor. Repeat this step to obtain a large number of image samples and their corresponding sensor readings. In this step, if it is inconvenient to install the sensor, the state estimation result corresponding to the highest resolution image can be used as the state variable reading x.

[0053] Step 1.3: Obtain the estimated state value of the control system corresponding to each image sample using the image processing algorithm selected in Step 1.1. And record the time τ required for the image processing process. p The system state estimate obtained from image processing The image processing error ω is obtained by subtracting the corresponding state variable reading x.

[0054] Step 1.4: Change the camera resolution r, and repeat steps 1.2 and 1.3 until all available camera resolutions have been traversed; obtain the functional relationship between all available resolutions and the upper bound of the modulus of the image processing error. The functional relationship between τ and image processing time p (r), where

[0055] Step 2: Calculate the control feedback matrix K and sampling trigger parameter matrix Φ that satisfy the system stability conditions, where the stability conditions are given by the Lyapunov method. In this embodiment, the system stability conditions are: given parameters 0 < μ < 1, λ ≥ 1, 0 < β. i <1,β i If λ < 1, there exists a matrix P. i >0,Q i ≥0,R i ≥0,Z 1i >0,Z 2i >0,M i ,S i N i For any The following inequalities hold:

[0056]

[0057] in

[0058]

[0059]

[0060]

[0061] Then when ω≡0, the system is globally uniformly asymptotically stable; when At that time, the range of the system's zero-state response lies within a series of elliptic x... T P i Within the intersection of x≤1, where It is the maximum possible value of the modulus of image processing error.

[0062] Step 3: Start the system online. (e.g., ...) Figure 3 As shown, the initial time is t = 0; the trigger time list and the corresponding resolution setting list are initialized; the camera is set to sample images at the minimum resolution at the initial time (t0 = 0), i.e., r(t0) = r min .

[0063] Step 4, at t=t k At time t, jump to step 5, where t kThis is the k-th trigger moment; otherwise, skip to step 6.

[0064] Step 5: The camera uses a resolution r(t) k The system samples the data and transmits the images to the image processing unit via a network; the image processing unit then processes the images to obtain an estimate of the control system's state. The data is sent to the local controller; the time delay from image sampling to the local controller receiving the state estimate is τ. k .

[0065] Step 6, at t=t k +τ k At that moment, the local controller received the state estimate. Skip to step 7; otherwise, skip to step 9.

[0066] Step 7: The feedback control module updates its state prediction. The calculation method is as follows: Where A is the system state matrix and B is the system input matrix.

[0067] Step 8: The sampling decision module updates the decisions regarding the sampling time and resolution configuration, such as... Figure 4 As shown, it includes the following steps:

[0068] Step 8.1, let s = t k +τ k ;

[0069] Step 8.2: Determine whether the system stability condition is met at time s. in It is the quadratic error between the current state and the state at the last sampling time (x(t) - x(t)). k )) T Φ(x(t)-x(t k The upper bound of )) It is the quadratic state μx at the last sampling time. T (t k )Φx(t k The lower bound of ) is given by μ, where μ is a pre-selected weight parameter, 0 < μ < 1. In this embodiment, the correlation function of the system stability condition is expressed as:

[0070]

[0071]

[0072]

[0073]

[0074]

[0075] If the system stability condition is not met, proceed to step 8.3; otherwise, proceed to step 8.4.

[0076] Step 8.3: For each possible image resolution r, calculate the corresponding total image transmission and processing delay. The calculation method is as follows: Where τ u =αr 2 / b represents the uplink image transmission delay, τ d For downlink state transmission delay estimation, h is the sampling interval of the controlled object (i.e., the duration between two consecutive times t and t+1), b is the network bandwidth, and αr 2 Let α represent the size of the image data, where α is a constant; let like Then let in h The minimum sampling time interval allowed by the camera; if Then let Calculate the stability boundary distance function C(r,t) corresponding to resolution r. k In this embodiment, the function is:

[0077]

[0078] Choose C(r,t) k Maximum resolution r * and its corresponding trigger time t k+1|k , will t k+1|k t k+1|k +1、……、 Add them to the trigger time list and set their corresponding image resolution to r. * Skip to step 9;

[0079] Step 8.4: If s or a time earlier than it is in the trigger time list, remove it and remove the corresponding resolution setting from the resolution setting list; let s = s + 1; if If so, proceed to step 8.2; otherwise, proceed to step 8.5. This is the maximum possible end-to-end delay;

[0080] Step 8.5, if t k +1、t k +2、……、 If none of them are in the trigger time list, then... Add it to the trigger time list and set the corresponding resolution to the minimum resolution.

[0081] Step 9: Update control commands using the feedback control module. The actuator executes control instructions; if t <T tmnl Then skip to step 4; otherwise, if t = T tmnl Then the operation will terminate, where T tmnl The pre-set termination time.

[0082] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. An adaptive image sampling method for visual perception control systems, characterized in that, Includes the following steps: Step 1: Model the relationship between processing error, processing delay and image resolution in visual perception tasks; Step 2: Determine the control feedback matrix K and sampling trigger parameter matrix Φ that satisfy the system stability conditions; Step 3: Start the system online operation; the initial time is t=0; Initialize the trigger time list and the corresponding resolution setting list; The camera is set to sample images at the initial time t0 = 0, with minimum resolution, i.e., r(t0) = r min ; Step 4, at t=t k At time t, jump to step 5, where t k This is the k-th trigger moment; otherwise, skip to step 6. Step 5: The camera uses a resolution r(t) k The system samples the images and transmits them to the image processing unit via a network; the image processing unit then processes the images to obtain an estimate of the control system's state. The data is sent to the local controller; the time delay from image sampling to the local controller receiving the state estimate is τ. k ; Step 6, at t=t k +τ k At that moment, the local controller receives the state estimate. Skip to step 7; otherwise, skip to step 9. Step 7: The feedback control module updates the state prediction. Step 8: The sampling decision module updates the decisions regarding the sampling time and resolution configuration; Step 9: Update control commands using the feedback control module. The actuator executes control instructions; if t < T tmnl Then skip to step 4; otherwise, if t = T tmnl Then the operation will terminate, where T tmnl The pre-set termination time.

2. The image adaptive sampling method for visual perception control systems as described in claim 1, characterized in that, Step 1 involves experimentally determining how the resolution of visual data affects image processing latency and state estimation error.

3. The image adaptive sampling method for visual perception control systems as described in claim 1, characterized in that, Step 1 includes the following steps: Step 1.1: Determine the camera settings and select the image processing algorithm. This allows the system to obtain the system status values ​​required for system control by processing the images obtained from the camera. Step 1.2: Set the camera resolution to a fixed value r and capture images. At the same time as the camera sampling, use a high-precision sensor to collect the target state variable value x. Use a unified timer to control the camera and the high-precision sensor. Step 1.3: Obtain the estimated state value of the control system corresponding to each image sample using the image processing algorithm selected in Step 1.

1. And record the time τ required for the image processing process. p The system state estimate obtained from image processing The image processing error ω is obtained by subtracting the corresponding state variable value x. Step 1.4: Change the camera resolution r, and repeat steps 1.2 and 1.3 until all available camera resolutions have been traversed; obtain the functional relationship between all available resolutions and the upper bound of the modulus of the image processing error. The functional relationship between τ and image processing time p (r), where 4. The image adaptive sampling method for visual perception control systems as described in claim 3, characterized in that, Step 1.1, camera settings include camera model, placement, parameter settings, and ambient light settings.

5. The image adaptive sampling method for visual perception control systems as described in claim 3, characterized in that, An alternative to step 1.2 is to not install a high-precision sensor and use the state estimation result corresponding to the highest resolution image as the state variable value x.

6. The image adaptive sampling method for visual perception control systems as described in claim 1, characterized in that, In step 2, the stability condition is given by the Lyapunov method.

7. The image adaptive sampling method for visual perception control systems as described in claim 1, characterized in that, Step 7, The calculation method is as follows: Where A is the system state matrix and B is the system input matrix.

8. The image adaptive sampling method for visual perception control systems as described in claim 1, characterized in that, In step 8, a stability boundary distance function is defined based on the system stability criterion, which is used as the objective function to dynamically adjust the visual information sampling strategy.

9. The image adaptive sampling method for visual perception control systems as described in claim 1, characterized in that, Step 8 includes the following steps: Step 8.1, let s = t k +τ k ; Step 8.2: Determine whether the system stability condition is met at time s. in It is the quadratic error between the current state and the state at the last sampling time (x(t) - x(t)). k )) T Φ(x(t)-x(t k The upper bound of )) It is the quadratic state μx at the last sampling time. T (t k )Φx(t k The lower bound of ) is μ, which is a pre-selected weight parameter, 0 < μ < 1; if the system stability condition is not met, jump to step 8.3; otherwise, jump to step 8.4; Step 8.3: For each image resolution r, calculate the corresponding total image transmission and processing delay. make like Then let in h The minimum sampling time interval allowed by the camera; if Then let Calculate the stability boundary distance function C(r, t) corresponding to resolution r. k ); Select C(r, t) k Maximum resolution r * and its corresponding trigger time t k+1|k , will t k+1|k t k+1|k +1、......、 Add it to the trigger time list and set the corresponding image resolution to r. * Skip to step 9; Step 8.4: If s or a previous time is in the trigger time list, remove it and remove the corresponding resolution setting from the resolution setting list; let s = s + 1; if If so, proceed to step 8.2; otherwise, proceed to step 8.

5. This is the maximum end-to-end delay; Step 8.5, if t k +1、t k +2、......、 If none of them are in the trigger time list, then... Add it to the trigger time list and set the corresponding resolution to the minimum resolution.

10. The image adaptive sampling method for a vision perception control system as described in claim 9, characterized in that, Step 8.3, The calculation method is as follows: Where τ u =αr 2 / b represents the uplink image transmission delay, τ d For downlink state transmission delay estimation, h is the sampling interval of the controlled object, b is the network bandwidth, and αr 2 This represents the size of the image data, where α is a constant.

Images