Dynamic observation capability quantitative evaluation method based on optical sensor

Abstract

The invention discloses a dynamic observation capability quantitative evaluation method based on an optical sensor, and relates to the technical field of automatic control, dynamic information of a workpiece is collected in real time through the optical sensor, the dynamic information comprises the position P, the speed V, the acceleration A and the ambient light intensity L, the data are summarized into a feature vector D1, and a complete trajectory prediction model is constructed. On the basis, a prediction error E is calculated by comparing a prediction position Pf and an actual position Pm, and the track model is dynamically adjusted by utilizing error feedback. The series of feedback mechanisms not only enhance the prediction ability of the model to the motion trail of the workpiece, but also effectively reduce the grabbing error. By combining the SNR and FOV factors of the optical sensor, the dynamic observation capability coefficient Cd of the optical sensor is calculated, the working action of the robot is further optimized, and the problem that the observation capability of the optical sensor is insufficient in the dynamic environment can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of automation control technology, and in particular to a quantitative evaluation method for dynamic observation capability based on an optical sensor. Background Art

[0002] The combination of automation technology and advanced sensor technology has promoted the automation and intelligent transformation of production lines. Optical sensors play a vital role in the collaborative application of industrial robots and machine vision technology. Optical sensors are commonly used sensing devices in modern intelligent manufacturing, which can quickly and accurately obtain information about the environment and the state of objects. In industrial automation, industrial robots need to perform complex tasks such as grasping, handling, and assembly.

[0003] In the dynamic environment of the production line, optical sensors monitor the position, speed and trajectory of the workpiece in real time, which is the core technology to ensure the precise operation of the robot. However, the current sensor system has some performance bottlenecks when facing high-speed moving workpieces. The main problems are the insufficient real-time capture capability of the sensor and the low accuracy of the prediction model. Especially when dealing with complex motion trajectories, the data captured by the sensor often has time delays or ambiguity, resulting in inaccurate predictions of the future position of the workpiece. In addition, the existing dynamic observation methods are mostly qualitative analysis, lacking quantitative evaluation and optimization of sensor capabilities, and cannot systematically reflect the performance of sensors in actual scenarios. This situation affects the robot's grasping efficiency of dynamic workpieces and limits the overall intelligence level of the production line. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a quantitative evaluation method for dynamic observation capability based on an optical sensor, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented by the following technical scheme: A quantitative evaluation method of dynamic observation capability based on an optical sensor, comprising the following steps: S1, collect real-time data of the workpieces on the production line through optical sensors, collect dynamic information of the workpieces, including position P, speed V, acceleration A and ambient light intensity L, and summarize them to obtain feature vector D1; S2, construct a trajectory prediction model based on the feature vector D1, use the trajectory prediction model to predict the motion path information within a fixed period based on the current state and historical trajectory of the workpiece, obtain the predicted position Pf, and integrate the predicted position Pf to form the feature vector D2; S3, using an optical sensor to measure the actual position Pm of the workpiece in real time and compare it with the feature vector D2 to obtain a prediction error E, and then performing secondary integration on the prediction error E to form a feature vector D3, and then performing feedback on the feature vector D3 to control the trajectory prediction model; S4, calculating according to the characteristic vector D3 in combination with the signal-to-noise ratio SNR of the optical sensor, the field of view FOV and the characteristic vector D1, to obtain the dynamic observation capability coefficient Cd of the optical sensor; S5. Build a comprehensive optimization model based on the predicted position Pf, the prediction error E and the dynamic observation capability coefficient Cd, and dynamically adjust the machine working action to obtain the comprehensive optimization evaluation result O; S6. Match the comprehensive optimization evaluation result O with the preset machine grasping effect threshold Othr to obtain the machine work action evaluation strategy scheme, and determine whether the machine continues to work and iterates the optimization according to the machine work action evaluation strategy scheme.

[0006] Preferably, said S1 includes S11 and S12; S11. Real-time data collection of workpieces on the production line is performed through optical sensors to collect dynamic information of the workpieces, including image processing measurement of the spatial position of the workpiece, obtaining the position coordinates of the X-axis, Y-axis and Z-axis, extracting the position P, continuously collecting data of the position P, calculating the instantaneous speed of the workpiece using time intervals, obtaining the speed V, calculating the instantaneous speed information of the workpiece using continuous speed V data, and obtaining the acceleration A; Among them, optical sensors are devices used to capture, detect and analyze optical information in objects and the environment. Optical sensors generate relevant physical data by collecting the reflection, refraction and scattering characteristics of light for calculation and dynamic observation, and then obtain the position P, velocity V, acceleration A and ambient light intensity L.

[0007] Preferably, S12, the light intensity information of the production line environment is simultaneously measured by an optical sensor, marked as ambient light intensity L, and the position P, speed V, acceleration A and ambient light intensity L are synchronously summarized to obtain a feature vector D1, wherein the feature vector D1 is specifically D1={P, V, A, L}.

[0008] Preferably, S2 includes S21 and S22; S21, constructing a trajectory prediction model according to the feature vector D1, using the trajectory prediction model to predict the motion path information within a fixed period based on the current state and historical trajectory of the workpiece, and obtaining the predicted position Pf; The predicted position Pf is obtained by the following calculation formula: ; Where Pf(t) represents the predicted position at time t, P(t0) represents the position of the workpiece at time t0, V(t0) represents the velocity of the workpiece at time t0, A(t0) represents the acceleration of the workpiece at time t0, and △t represents the predicted time step, which specifically represents the time change starting from the current time t0.

[0009] S22. The predicted positions Pf obtained by calculation are integrated to form a new feature vector D2. The feature vector D2 is specifically D2={Pf(t1), Pf(t2), ..., Pf(tn)}.

[0010] Preferably, said S3 includes S31 and S32; S31, measuring the position Pm of the workpiece at time ti by an optical sensor, and comparing the position Pm with the predicted position Pf predicted in the trajectory prediction model, obtaining the error E at time ti, which highlights the deviation state between the actual and predicted positions; The error E is calculated by using the Euclidean distance formula: ; Where Pf,x(ti), Pf,y(ti) and Pf,z(ti) represent the x, y and z axis coordinate values ​​of the predicted position Pf at time ti, respectively, and Pm,x(ti), Pm,y(ti) and Pm,z(ti) represent the x, y and z axis coordinate values ​​of the position Pm at time ti, respectively.

[0011] Preferably, S32, integrating the errors E at different times ti to generate a feature vector D3, and feeding the feature vector D3 back to the trajectory prediction model for regulation, and continuously correcting the future predicted position Pf through the feature vector D3, wherein the feature vector D3 is specifically D3={E(t1), E(t2), ..., E(tn)}; Among them, the correction of the future predicted position Pf is specifically regulated by the following calculation formula: ; Where Pf,new(ti+1) represents the adjusted new predicted position Pf for the next time ti+1, and Pf(ti+1) represents the predicted position at time ti+1. Represents the feedback adjustment coefficient, which is specifically used to correct the impact of error correction on the predicted position.

[0012] Preferably, said S4 includes S41 and S42; S41, extracting parameters for evaluating the dynamic observation capability of the optical sensor, including the signal-to-noise ratio SNR and the field of view FOV, wherein the signal-to-noise ratio SNR is obtained by calculating the ratio of the signal intensity Ssig and the noise intensity Snoise detected by the optical sensor, specifically by The ratio calculation formula is obtained; the field of view FOV is obtained by extracting the spatial range that the optical sensor can observe, which specifically represents the range angle that the optical sensor can monitor.

[0013] Preferably, S42, dynamically obtaining the observation capability of the optical sensor in a dynamic environment by fitting the light intensity L in the feature vector D1 and the error E in the feature vector D3 with the signal-to-noise ratio SNR and the field of view FOV, and obtaining a dynamic observation capability coefficient Cd; The dynamic observation capability coefficient Cd is The calculation formula is obtained, where n represents the number of time points, i represents the index of time t, and specifically represents time ti.

[0014] Preferably, a comprehensive optimization model is constructed according to the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, and a comprehensive optimization evaluation result O is obtained by combining the input information of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd to dynamically adjust the robot working action; The comprehensive optimization evaluation result O is obtained by the following calculation formula: ; Where o1, o2 and o3 represent the preset weight values ​​of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, respectively, and , the specific value is set by the user.

[0015] Preferably, the comprehensive optimization evaluation result O is matched with the preset machine grasping effect threshold Othr to determine whether the current machine working action meets the requirements, obtain the machine working action evaluation strategy scheme, and determine whether the machine continues to work and iterates the optimization according to the machine working action evaluation strategy scheme; The machine work action evaluation strategy scheme is obtained by the following matching method: When the comprehensive optimization evaluation result O≥machine grasping effect threshold Othr, the content of the machine work action evaluation strategy solution is qualified, and the robot continues to perform the current grasping action without making any adjustments and executes the preset workflow; When the comprehensive optimization evaluation result O is less than the machine grasping effect threshold Othr, the content of the machine work action evaluation strategy solution is obtained as unqualified, triggering the iterative optimization process to adjust the machine action, including adjusting the error E, dynamic observation capability Cd and predicted position Pf.

[0016] The present invention provides a quantitative evaluation method for dynamic observation capability based on an optical sensor, which has the following beneficial effects: (1) Through steps S1 to S6, the grasping accuracy and efficiency of the robot in the production line are effectively improved. First, the system collects the dynamic information of the workpiece in real time through the optical sensor, including the position P, velocity V, acceleration A and ambient light intensity L, and summarizes these data into the feature vector D1 to build a complete trajectory prediction model. On this basis, by comparing the predicted position Pf and the actual position Pm, the prediction error E is calculated, and these error feedbacks are used to dynamically adjust the trajectory model. This series of feedback mechanisms not only enhances the model's ability to predict the motion trajectory of the workpiece, but also effectively reduces the grasping error. By combining the signal-to-noise ratio SNR and the field of view FOV of the optical sensor, the dynamic observation capability coefficient Cd of the optical sensor is calculated, which further optimizes the robot's working action and can effectively solve the problem of insufficient observation capability of the optical sensor in a dynamic environment, especially in the trajectory prediction, error feedback and real-time response of the high-speed workpiece. Through quantitative evaluation of dynamic observation capability, the system can greatly improve the robot's grasping success rate, reduce misoperation, and improve the intelligence level of the entire production line. At the same time, it overcomes the shortcomings of previous methods in prediction accuracy, system robustness and error control.

[0017] (2) By continuously collecting position P data, the velocity V and acceleration A of the workpiece are calculated, and a comprehensive analysis is performed in combination with the ambient light intensity L to generate a feature vector D1, ensuring that the dynamic state of the workpiece is fully captured. Subsequently, the trajectory prediction model constructed by the feature vector D1 can accurately predict the movement path of the workpiece at different times in the future and generate a predicted position Pf. By integrating the predicted position Pf, a new feature vector D2 is formed, which provides a reliable data basis for subsequent action optimization and effectively improves the prediction accuracy and response speed of the system. Especially in complex environments, the optical sensor can not only capture the fine movement state of the workpiece, but also dynamically analyze its future trajectory to ensure that the robot can adjust its movements faster and more accurately, thereby improving the efficiency and stability of the production line. This method combined with the ambient light intensity L further enhances the system's adaptability to external conditions such as lighting changes, ensuring the reliability of dynamic monitoring and prediction in a changing environment.

[0018] (3) By combining the error feedback of the actual and predicted positions Pf with the performance evaluation of the optical sensor, the automatic adjustment and self-adaptation capabilities of the system in a dynamic environment are significantly enhanced. By comparing the actual position Pm with the predicted position Pf, the error E is accurately calculated using the Euclidean distance formula, and the error at different times is fed back to the trajectory prediction model through the feature vector D3 to dynamically correct the future predicted position. This adjustment mechanism based on error feedback can effectively reduce the prediction deviation and ensure the continuity and accuracy of the workpiece motion trajectory prediction. Secondly, by extracting key parameters such as the signal-to-noise ratio SNR and field of view FOV of the optical sensor, and combining the ambient light intensity L and the error E, the dynamic observation capability coefficient Cd of the optical sensor is dynamically calculated. It can comprehensively evaluate the performance of the optical sensor under different lighting and noise conditions, enabling the system to adaptively adjust the observation strategy and improve the observation accuracy. It effectively solves the defect of unstable observation capability of the traditional system under the fluctuation of optical sensor performance and changes in the external environment, improves the reliability and accuracy of the sensor in a complex production environment, and ensures that the robot's working actions can be executed efficiently in a changing environment.

[0019] (4) By constructing a comprehensive optimization model and combining the multi-dimensional input information of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, the precise adjustment and real-time optimization of the robot's working action are ensured. In this process, the system calculates the comprehensive optimization evaluation result O, matches it according to the preset machine grasping effect threshold Othr, and automatically generates a machine working action evaluation strategy. It also automatically triggers iterative optimization and dynamically adjusts the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, thereby improving the robot's grasping accuracy and execution efficiency, and improving the overall operation efficiency and grasping success rate of the production line. Especially when the machine's working conditions change, this method can actively optimize the robot's motion control, ensure the stability of the system under dynamic conditions, reduce operational errors caused by excessive errors or sensor performance fluctuations, and improve the system's adaptability and intelligence level. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The figure is a schematic diagram of the steps of a method for quantitatively evaluating dynamic observation capability based on an optical sensor of the present invention. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Example 1

[0022] The present invention provides a quantitative evaluation method for dynamic observation capability based on optical sensors. Figure 1 , including the following steps: S1, collect real-time data of the workpieces on the production line through optical sensors, collect dynamic information of the workpieces, including position P, speed V, acceleration A and ambient light intensity L, and summarize them to obtain feature vector D1; S2, construct a trajectory prediction model based on the feature vector D1, use the trajectory prediction model to predict the motion path information within a fixed period based on the current state and historical trajectory of the workpiece, obtain the predicted position Pf, and integrate the predicted position Pf to form the feature vector D2; S3, using an optical sensor to measure the actual position Pm of the workpiece in real time and compare it with the feature vector D2 to obtain a prediction error E, and then performing secondary integration on the prediction error E to form a feature vector D3, and then performing feedback on the feature vector D3 to control the trajectory prediction model; S4, calculating according to the characteristic vector D3 in combination with the signal-to-noise ratio SNR of the optical sensor, the field of view FOV and the characteristic vector D1, to obtain the dynamic observation capability coefficient Cd of the optical sensor; S5. Build a comprehensive optimization model based on the predicted position Pf, the prediction error E and the dynamic observation capability coefficient Cd, and dynamically adjust the machine working action to obtain the comprehensive optimization evaluation result O; S6. Match the comprehensive optimization evaluation result O with the preset machine grasping effect threshold Othr to obtain the machine work action evaluation strategy scheme, and determine whether the machine continues to work and iterates the optimization according to the machine work action evaluation strategy scheme.

[0023] In this embodiment, the steps S1 to S6 are used to effectively improve the grasping accuracy and efficiency of the robot in the production line. First, the system collects the dynamic information of the workpiece in real time through the optical sensor, including the position P, speed V, acceleration A and ambient light intensity L, and summarizes these data into the feature vector D1 to build a complete trajectory prediction model. On this basis, by comparing the predicted position Pf and the actual position Pm, the prediction error E is calculated, and the trajectory model is dynamically adjusted using these error feedbacks. This series of feedback mechanisms not only enhances the model's ability to predict the motion trajectory of the workpiece, but also effectively reduces the grasping error. By combining the signal-to-noise ratio SNR and the field of view FOV factors of the optical sensor, the dynamic observation capability coefficient Cd of the optical sensor is calculated, which further optimizes the robot's working action and can effectively solve the problem of insufficient observation capability of the optical sensor in a dynamic environment, especially in the trajectory prediction of high-speed workpieces, error feedback and real-time response of the system. Through the quantitative evaluation of the dynamic observation capability, the system can greatly improve the robot's grasping success rate, reduce misoperation, and improve the intelligence level of the entire production line, while overcoming the shortcomings of previous means in prediction accuracy, system robustness and error control. Example 2

[0024] This embodiment is explained in Example 1, please refer to Figure 1 , specifically: the S1 includes S11 and S12; S11. Real-time data collection of workpieces on the production line is performed through optical sensors to collect dynamic information of the workpieces, including image processing measurement of the spatial position of the workpiece, obtaining the position coordinates of the X-axis, Y-axis and Z-axis, extracting the position P, continuously collecting data of the position P, calculating the instantaneous speed of the workpiece using time intervals, obtaining the speed V, calculating the instantaneous speed information of the workpiece using continuous speed V data, and obtaining the acceleration A; Among them, optical sensors are devices used to capture, detect and analyze optical information in objects and the environment. Optical sensors generate relevant physical data by collecting the reflection, refraction and scattering characteristics of light for calculation and dynamic observation, and then obtain the position P, velocity V, acceleration A and ambient light intensity L.

[0025] S12. The light intensity information of the production line environment is measured simultaneously by an optical sensor, marked as ambient light intensity L, and the position P, speed V, acceleration A and ambient light intensity L are synchronously summarized to obtain a feature vector D1, wherein the feature vector D1 is specifically D1={P, V, A, L}.

[0026] The S2 includes S21 and S22; S21, constructing a trajectory prediction model according to the feature vector D1, using the trajectory prediction model to predict the motion path information within a fixed period based on the current state and historical trajectory of the workpiece, and obtaining the predicted position Pf; The predicted position Pf is obtained by the following calculation formula: ; Where Pf(t) represents the predicted position at time t, P(t0) represents the position of the workpiece at time t0, V(t0) represents the velocity of the workpiece at time t0, A(t0) represents the acceleration of the workpiece at time t0, and △t represents the predicted time step, which specifically represents the time change starting from the current time t0.

[0027] S22. The predicted positions Pf obtained by calculation are integrated to form a new feature vector D2. The feature vector D2 is specifically D2={Pf(t1), Pf(t2), ..., Pf(tn)}.

[0028] In this embodiment, by continuously collecting position P data, the speed V and acceleration A of the workpiece are calculated, and a comprehensive analysis is performed in combination with the ambient light intensity L to generate a feature vector D1 to ensure that the dynamic state of the workpiece is fully captured. Subsequently, through the trajectory prediction model constructed by the feature vector D1, the movement path of the workpiece at different moments in the future can be accurately predicted to generate a predicted position Pf. By integrating the predicted position Pf, a new feature vector D2 is formed, which provides a reliable data basis for subsequent action optimization and effectively improves the prediction accuracy and response speed of the system. Especially in complex environments, the optical sensor can not only capture the fine movement state of the workpiece, but also dynamically analyze its future trajectory to ensure that the robot can adjust the action faster and more accurately, thereby improving the efficiency and stability of the production line. This method of combining the ambient light intensity L further enhances the system's adaptability to external conditions such as changes in lighting, and ensures the reliability of dynamic monitoring and prediction in a changing environment. Example 3

[0029] This embodiment is explained in Example 1, please refer to Figure 1 , specifically: S3 includes S31 and S32; S31, measuring the position Pm of the workpiece at time ti by an optical sensor, and comparing the position Pm with the predicted position Pf predicted in the trajectory prediction model, obtaining the error E at time ti, which highlights the deviation state between the actual and predicted positions; The error E is calculated by using the Euclidean distance formula: ; Where Pf,x(ti), Pf,y(ti) and Pf,z(ti) represent the x, y and z axis coordinate values ​​of the predicted position Pf at time ti, respectively, and Pm,x(ti), Pm,y(ti) and Pm,z(ti) represent the x, y and z axis coordinate values ​​of the position Pm at time ti, respectively.

[0030] S32, integrating the errors E at different times ti to generate a feature vector D3, and feeding the feature vector D3 back to the trajectory prediction model for regulation, and continuously correcting the future predicted position Pf through the feature vector D3, wherein the feature vector D3 is specifically D3={E(t1), E(t2), ..., E(tn)}; Among them, the correction of the future predicted position Pf is specifically regulated by the following calculation formula: ; Where Pf,new(ti+1) represents the adjusted new predicted position Pf for the next time ti+1, and Pf(ti+1) represents the predicted position at time ti+1. Represents the feedback adjustment coefficient, which is specifically used to correct the impact of error correction on the predicted position.

[0031] The S4 includes S41 and S42; S41, extracting parameters for evaluating the dynamic observation capability of the optical sensor, including the signal-to-noise ratio SNR and the field of view FOV, wherein the signal-to-noise ratio SNR is obtained by calculating the ratio of the signal intensity Ssig and the noise intensity Snoise detected by the optical sensor, specifically by The ratio calculation formula is obtained; the field of view FOV is obtained by extracting the spatial range that the optical sensor can observe, which specifically represents the range angle that the optical sensor can monitor.

[0032] S42, dynamically obtain the observation capability of the optical sensor in a dynamic environment by fitting the light intensity L in the feature vector D1 and the error E in the feature vector D3 with the signal-to-noise ratio SNR and the field of view FOV, and obtain the dynamic observation capability coefficient Cd; The dynamic observation capability coefficient Cd is The calculation formula is obtained, where n represents the number of time points, i represents the index of time t, and specifically represents time ti.

[0033] In this embodiment, by combining the error feedback of the actual and predicted positions Pf with the performance evaluation of the optical sensor, the automatic adjustment and self-adaptation capabilities of the system in a dynamic environment are significantly enhanced. By comparing the actual position Pm and the predicted position Pf, the error E is accurately calculated using the Euclidean distance formula, and the error at different times is fed back to the trajectory prediction model through the characteristic vector D3 to dynamically correct the predicted position in the future. This adjustment mechanism based on error feedback can effectively reduce the prediction deviation and ensure the continuity and accuracy of the workpiece motion trajectory prediction. Secondly, by extracting key parameters such as the signal-to-noise ratio SNR and the field of view FOV of the optical sensor, and combining the ambient light intensity L and the error E, the dynamic observation capability coefficient Cd of the optical sensor is dynamically calculated. It can comprehensively evaluate the performance of the optical sensor under different lighting and noise conditions, so that the system can adaptively adjust the observation strategy and improve the observation accuracy, effectively solving the defect of unstable observation capability of the traditional system under the fluctuation of optical sensor performance and changes in the external environment, improving the reliability and accuracy of the sensor in a complex production environment, and ensuring that the robot's working actions can be performed efficiently in a changing environment. Example 4

[0034] This embodiment is explained in Example 1, please refer to Figure 1 Specifically: wherein, a comprehensive optimization model is constructed according to the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, and the comprehensive optimization evaluation result O is obtained by combining the input information of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd to dynamically adjust the robot working action; The comprehensive optimization evaluation result O is obtained by the following calculation formula: ; Where o1, o2 and o3 represent the preset weight values ​​of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, respectively, and , the specific value is set by the user.

[0035] According to the comprehensive optimization evaluation result O, the preset machine grasping effect threshold Othr is matched to determine whether the current machine working action meets the requirements, obtain the machine working action evaluation strategy plan, and determine whether the machine continues to work and iterate optimization according to the machine working action evaluation strategy plan; The machine work action evaluation strategy scheme is obtained by the following matching method: When the comprehensive optimization evaluation result O≥machine grasping effect threshold Othr, the content of the machine work action evaluation strategy solution is qualified, and the robot continues to perform the current grasping action without making any adjustments and executes the preset workflow; When the comprehensive optimization evaluation result O is less than the machine grasping effect threshold Othr, the content of the machine work action evaluation strategy solution is obtained as unqualified, triggering the iterative optimization process to adjust the machine action, including adjusting the error E, dynamic observation capability Cd and predicted position Pf.

[0036] In this embodiment, by constructing a comprehensive optimization model and combining the multi-dimensional input information of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, the precise adjustment and real-time optimization of the robot's working action are ensured. In this process, the system automatically generates a machine working action evaluation strategy by calculating the comprehensive optimization evaluation result O and matching it according to the preset machine grasping effect threshold Othr. When O reaches or exceeds the threshold Othr, the robot can perform the current grasping action without adjustment to ensure a smooth workflow; when O is less than the threshold, the system triggers iterative optimization and dynamically adjusts the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, thereby improving the robot's grasping accuracy and execution efficiency, and improving the overall operating efficiency and grasping success rate of the production line. Especially when the machine working conditions change, this method can actively optimize the robot's motion control, ensure the stability of the system under dynamic conditions, reduce operational errors caused by excessive errors or sensor performance fluctuations, and improve the system's adaptive ability and intelligence level.

[0037] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A quantitative evaluation method for dynamic observation capability based on optical sensors, characterized in that: The following steps are involved: S1, collect real-time data of the workpieces on the production line through optical sensors, collect dynamic information of the workpieces, including position P, speed V, acceleration A and ambient light intensity L, and summarize them to obtain feature vector D1; S2. Construct a trajectory prediction model based on the feature vector D1, use the trajectory prediction model to predict the motion path information within a fixed period based on the current state and historical trajectory of the workpiece, obtain the predicted position Pf, and integrate the predicted position Pf. Form feature vector D2; S3, measure the actual position Pm of the workpiece in real time through the optical sensor and compare it with the feature vector D2, obtain the prediction error E, and perform secondary integration on the prediction error E, The feature vector D3 is formed, and the feature vector D3 is fed back to control the trajectory prediction model; S4, calculating according to the characteristic vector D3 in combination with the signal-to-noise ratio SNR of the optical sensor, the field of view FOV and the characteristic vector D1, to obtain the dynamic observation capability coefficient Cd of the optical sensor; S5. Build a comprehensive optimization model based on the predicted position Pf, the prediction error E and the dynamic observation capability coefficient Cd, and dynamically adjust the machine working action to obtain the comprehensive optimization evaluation result O; S6. Match the comprehensive optimization evaluation result O with the preset machine grasping effect threshold Othr to obtain the machine work action evaluation strategy scheme, and determine whether the machine continues to work and iterates the optimization according to the machine work action evaluation strategy scheme.

2. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 1, characterized in that: Said S1 includes S11 and S12; S11. Real-time data collection of workpieces on the production line is performed through optical sensors to collect dynamic information of the workpieces, including image processing measurement of the spatial position of the workpiece to obtain the position coordinates of the X-axis, Y-axis and Z-axis, extract the position P, continuously collect data of the position P, calculate the instantaneous speed of the workpiece using the time interval, obtain the speed V, calculate the instantaneous speed information of the workpiece using the continuous speed V data, and obtain the acceleration A; Among them, optical sensors are devices used to capture, detect and analyze optical information in objects and the environment. Optical sensors generate relevant physical data by collecting the reflection, refraction and scattering characteristics of light for calculation and dynamic observation, and then obtain the position P, velocity V, acceleration A and ambient light intensity L.

3. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 2, characterized in that: S12. The light intensity information of the production line environment is measured simultaneously by an optical sensor, marked as ambient light intensity L, and the position P, speed V, acceleration A and ambient light intensity L are synchronously summarized to obtain a feature vector D1, wherein the feature vector D1 is specifically D1={P, V, A, L}.

4. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 3, characterized in that: The S2 includes S21 and S22; S21, construct a trajectory prediction model according to the feature vector D1, and use the trajectory prediction model to predict the motion path information within a fixed period of the current state and historical trajectory of the workpiece to obtain the predicted position Pf; the predicted position Pf is obtained by the following calculation formula: ; Where Pf(t) represents the predicted position at time t, P(t0) represents the position of the workpiece at time t0, V(t0) represents the velocity of the workpiece at time t0, A(t0) represents the acceleration of the workpiece at time t0, and △t represents the predicted time step, which specifically represents the time change starting from the current time t0; S22. The predicted positions Pf obtained by calculation are integrated to form a new feature vector D2. The feature vector D2 is specifically D2={Pf(t1), Pf(t2), ..., Pf(tn)}.

5. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 1, characterized in that: The S3 includes S31 and S32; S31, measuring the position Pm of the workpiece at time ti by an optical sensor, and comparing the position Pm with the predicted position Pf predicted in the trajectory prediction model, obtaining the error E at time ti, which highlights the deviation state between the actual and predicted positions; The error E is calculated by using the Euclidean distance formula: ; Where Pf,x(ti), Pf,y(ti) and Pf,z(ti) represent the x, y and z axis coordinate values ​​of the predicted position Pf at time ti, respectively, and Pm,x(ti), Pm,y(ti) and Pm,z(ti) represent the x, y and z axis coordinate values ​​of the position Pm at time ti, respectively.

6. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 5, characterized in that: S32, integrating the errors E at different times ti to generate a feature vector D3, and feeding the feature vector D3 back to the trajectory prediction model for regulation, and continuously correcting the future predicted position Pf through the feature vector D3, wherein the feature vector D3 is specifically D3={E(t1), E(t2), ..., E(tn)}; Among them, the correction of the future predicted position Pf is specifically regulated by the following calculation formula: ; In the formula, Pf,new(ti+1) represents the adjusted new predicted position Pf for the next time ti+1, and Pf(ti+1) represents the predicted position at time ti+1. Represents the feedback adjustment coefficient, which is specifically used to correct the impact of error correction on the predicted position.

7. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 1, characterized in that: The S4 includes S41 and S42; S41, extracting parameters for evaluating the dynamic observation capability of the optical sensor, including the signal-to-noise ratio SNR and the field of view FOV, wherein the signal-to-noise ratio SNR is obtained by calculating the ratio of the signal intensity Ssig and the noise intensity Snoise detected by the optical sensor, specifically by The ratio calculation formula is obtained; the field of view FOV is obtained by extracting the spatial range that the optical sensor can observe, which specifically represents the range angle that the optical sensor can monitor.

8. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 7, characterized in that: S42, dynamically obtain the observation capability of the optical sensor in a dynamic environment by fitting the light intensity L in the feature vector D1 and the error E in the feature vector D3 with the signal-to-noise ratio SNR and the field of view FOV, and obtain the dynamic observation capability coefficient Cd; The dynamic observation capability coefficient Cd is The calculation formula is obtained, where n represents the number of time points, i represents the index of time t, and specifically represents time ti.

9. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 8, characterized in that: A comprehensive optimization model is constructed based on the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd. By combining the input information of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, a comprehensive optimization evaluation result O is obtained to dynamically adjust the robot's working action. The comprehensive optimization evaluation result O is obtained by the following calculation formula: ; Where o1, o2 and o3 represent the preset weight values ​​of the predicted position Pf, the predicted error E and the dynamic observation capability coefficient Cd, respectively, and , the specific value is set by the user.

10. The method for quantitatively evaluating dynamic observation capability based on an optical sensor according to claim 9, characterized in that: According to the comprehensive optimization evaluation result O, the preset machine grasping effect threshold Othr is matched to determine whether the current machine working action meets the requirements, obtain the machine working action evaluation strategy plan, and determine whether the machine continues to work and iterate optimization according to the machine working action evaluation strategy plan; The machine work action evaluation strategy scheme is obtained by the following matching method: When the comprehensive optimization evaluation result O≥machine grasping effect threshold Othr, the content of the machine work action evaluation strategy solution is qualified, and the robot continues to perform the current grasping action without making any adjustments and executes the preset workflow; When the comprehensive optimization evaluation result O is less than the machine grasping effect threshold Othr, the content of the machine work action evaluation strategy solution is obtained as unqualified, triggering the iterative optimization process to adjust the machine action, including adjusting the error E, dynamic observation capability Cd and predicted position Pf.

Images