An operation monitoring method based on industrial robots

Abstract

This invention discloses an operation monitoring method for an industrial robot, belonging to the field of industrial robot monitoring technology. The method comprises the following steps: S1. capturing a real-life image in front of the industrial robot and calculating a stability coefficient based on two sample sets of the real-life image; S2. calculating a symbol value for each pixel point based on the stability coefficient of the real-life image; and S3. determining the obstacle area in front of the industrial robot based on the symbol value of each pixel point in the real-life image. This invention provides a clear target for obstacle avoidance and path planning for the industrial robot, facilitating subsequent control decisions.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial robot monitoring, and in particular relates to an operation monitoring method based on an industrial robot. Background Art

[0002] With the rapid development of industrial automation and intelligent manufacturing, industrial robots are increasingly used on production lines. While performing their tasks, industrial robots need to perceive their surroundings in real time, particularly identifying potential obstacles, to ensure safe and stable operation. Traditional obstacle detection methods primarily rely on hardware devices such as lidar, ultrasonic sensors, or infrared sensors. However, these methods suffer from high costs, complex deployment, and poor environmental adaptability.

[0003] In recent years, vision-based obstacle detection technology has become a research hotspot in the field of environmental perception for industrial robots due to its low cost, rich information, and ease of deployment. By analyzing images of the real scene in front of the industrial robot, the location and range of obstacles can be effectively identified. However, the complexity of real-world images (such as lighting variations, background interference, and object occlusion) poses challenges to obstacle detection. Therefore, extracting robust features from images and accurately determining obstacle areas is a key issue that needs to be addressed. Summary of the Invention

[0004] In order to solve the above problems, the present invention proposes an operation monitoring method based on an industrial robot.

[0005] The technical solution of the present invention is: an operation monitoring method based on an industrial robot comprises the following steps:

[0006] S1. Collect real-scene images in front of the industrial robot body and calculate the stability coefficient based on two sample sets of the real-scene images;

[0007] S2. Calculate the symbol value for each pixel based on the stability coefficient of the real scene image;

[0008] S3. Determine the obstacle area in front of the industrial robot body according to the symbol value of each pixel point in the real scene image.

[0009] Furthermore, S1 includes the following sub-steps:

[0010] S11, collecting real-scene images in front of the industrial robot body;

[0011] S12, extracting a positive sample set and a negative sample set of the real scene image;

[0012] S13. Calculate a stability coefficient based on the positive sample set and the negative sample set of the real scene image.

[0013] The beneficial effect of this further solution is that, in this invention, image pixels are divided into positive and negative sample sets, balancing the number of positive and negative samples, avoiding calculation bias caused by large sample size differences, and ensuring the accuracy of the stability coefficient calculation. The stability coefficient quantifies the linear correlation between the pixel values ​​of the positive and negative sample sets and the overall mean, reflecting the stability of the image region. This coefficient can be used to assess the consistency or difference of features in different image regions, providing an important statistical basis for obstacle detection.

[0014] Furthermore, in S12, the mean pixel value of all pixels in the real scene image is extracted, and the pixel points with pixel values ​​greater than the mean pixel value are taken as the positive sample set, and the pixel points with pixel values ​​less than or equal to the mean pixel value are taken as the negative sample set.

[0015] Furthermore, S13 includes the following sub-steps:

[0016] S131, taking the minimum value between the number of samples in the positive sample set and the number of samples in the negative sample set as the sample threshold value N;

[0017] S132, sorting the pixel values ​​of the positive sample set and the negative sample set from large to small, and extracting the top N pixel points in the positive sample set and the top N pixel points in the negative sample set;

[0018] S133. Calculate the stability coefficient of the real scene image based on the top N pixels in the positive sample set and the top N pixels in the negative sample set.

[0019] The beneficial effect of this further solution is that, by sorting and extracting the top-ranked pixels, the present invention focuses on key pixels with large pixel value differences. These pixels are more important for reflecting the characteristic differences of the image, which helps to improve the sensitivity of the stability coefficient to distinguishing obstacles from background. Calculating the stability coefficient using these selected key pixels can more accurately reflect the linear correlation between the pixel values ​​of the positive and negative sample sets and the overall mean deviation, thereby improving the accuracy of the stability coefficient in quantifying image stability.

[0020] Furthermore, in S133, the calculation formula of the stability coefficient w of the real scene image is: Where x n+ Indicates the pixel value of the nth pixel in the positive sample set, x n- Represents the pixel value of the nth pixel in the negative sample set, N represents the sample threshold value, Represents the mean pixel value of all pixels in the real scene image.

[0021] By using correlation analysis of pixel values, we can reduce the impact of factors such as illumination and noise on detection results and improve the robustness of the algorithm in complex environments.

[0022] Furthermore, S2 includes the following sub-steps:

[0023] S21. Generate a stability weight of the real scene image based on the stability coefficient;

[0024] S22. Generate a symbol value for each pixel according to the stability weight of the real scene image.

[0025] The beneficial effect of this further approach is that, in this invention, stable weights are generated in a random manner, introducing a degree of randomness to each real-world image. This allows the algorithm to produce diverse weight distributions when processing different images, thereby better adapting to various complex industrial scenarios. Different images may have different feature distributions, and random weights can prevent the algorithm from falling into a fixed pattern, improving its ability to capture diverse image features.

[0026] When calculating the random symbol value for a pixel, the average pixel value of its four neighboring pixels is used. This incorporates the pixel's local neighborhood information and considers the pixel's context within the image. This neighborhood information provides more clues to image features, allowing the generated symbol value to better reflect the pixel's local characteristics, helping to more accurately identify obstacle areas.

[0027] Furthermore, in S21, the calculation formula of the stability weight W is: Where w represents the stability coefficient of the real image, and rand(·) represents the random function. This function generates a random number between w and -w.

[0028] Furthermore, in S22, the calculation formula of the random symbol value F of the pixel point is: ;W represents the stability weight of the pixel, x represents the pixel value of the pixel, It represents the mean pixel value of the four neighboring pixels around the pixel point, and sign represents the sign function.

[0029] If necessary, the pixel values ​​can be de-dimensionalized to make the calculation reasonable.

[0030] Furthermore, in S3, the pixel points with a symbol value of -1 are regarded as the obstacle area in front of the industrial robot body.

[0031] The beneficial effect of the above-mentioned further solution is that, in the present invention, the stability of different image regions is quantified through the calculation of stability coefficients and stability weights, and pixels are classified and weighted. Pixels with a symbol value of -1 have distinct characteristics distinct from the background area, and identifying them as obstacle areas can highlight the characteristics of the obstacles. By determining the obstacle area in front of the industrial robot based on the symbol values ​​of the pixels, the abstract stability coefficient and symbol value can be converted into specific obstacle area information, providing a clear and specific target area for the industrial robot's obstacle avoidance and path planning, effectively ensuring the safe operation of the industrial robot.

[0032] The beneficial effects of the present invention are:

[0033] (1) The present invention divides image pixels into positive sample sets and negative sample sets. The calculation of the stability coefficient can quantify the linear correlation between the pixel values ​​of the positive and negative sample sets and the overall mean deviation, which helps to distinguish obstacle areas;

[0034] (2) The present invention calculates the symbol value for each pixel based on the stability coefficient, realizes the personalized analysis of each pixel in the image, can fully consider the characteristics and position of each pixel in the image, and improves the precision of obstacle detection;

[0035] (3) The present invention can quickly locate the obstacle area through the symbol value, because the symbol value has already classified the pixel points. It only needs to filter out the pixel points corresponding to the symbol value that meets the obstacle characteristics to determine the obstacle area, which improves the detection efficiency and provides a clear target for the obstacle avoidance and path planning of the industrial robot, facilitating subsequent control decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 Flowchart of the operation monitoring method of an industrial robot. DETAILED DESCRIPTION

[0037] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0038] like Figure 1 As shown, the present invention provides an operation monitoring method based on an industrial robot, comprising the following steps:

[0039] S1. Collect real-scene images in front of the industrial robot body and calculate the stability coefficient based on two sample sets of the real-scene images;

[0040] S2. Calculate the symbol value for each pixel based on the stability coefficient of the real scene image;

[0041] S3. Determine the obstacle area in front of the industrial robot body according to the symbol value of each pixel point in the real scene image.

[0042] In this embodiment of the present invention, S1 includes the following sub-steps:

[0043] S11, collecting real-scene images in front of the industrial robot body;

[0044] S12, extracting a positive sample set and a negative sample set of the real scene image;

[0045] S13. Calculate a stability coefficient based on the positive sample set and the negative sample set of the real scene image.

[0046] In this paper, image pixels are divided into positive and negative sample sets, balancing the number of positive and negative samples to avoid calculation bias caused by large sample size differences and ensure the accuracy of the stability coefficient calculation. The stability coefficient quantifies the linear correlation between the pixel values ​​of the positive and negative sample sets and the overall mean deviation, reflecting the stability of the image region. This coefficient can be used to assess the consistency or difference of features in different image regions, providing an important statistical basis for obstacle detection.

[0047] In an embodiment of the present invention, in S12, the mean pixel value of all pixels in the real scene image is extracted, and the pixel points with pixel values ​​greater than the mean pixel value are taken as the positive sample set, and the pixel points with pixel values ​​less than or equal to the mean pixel value are taken as the negative sample set.

[0048] In this embodiment of the present invention, S13 includes the following sub-steps:

[0049] S131, taking the minimum value between the number of samples in the positive sample set and the number of samples in the negative sample set as the sample threshold value N;

[0050] S132, sorting the pixel values ​​of the positive sample set and the negative sample set from large to small, and extracting the top N pixel points in the positive sample set and the top N pixel points in the negative sample set;

[0051] S133. Calculate the stability coefficient of the real scene image based on the top N pixels in the positive sample set and the top N pixels in the negative sample set.

[0052] In this method, by sorting and extracting the top-ranked pixels, we focus on key pixels with large pixel value differences. These pixels are more important for reflecting the characteristic differences of the image, helping to improve the stability coefficient's sensitivity to distinguishing obstacles from background. Using these selected key pixels to calculate the stability coefficient can more accurately reflect the linear correlation between the pixel values ​​of the positive and negative sample sets and the overall mean deviation, improving the stability coefficient's quantification accuracy for image stability.

[0053] In the embodiment of the present invention, in S133, the calculation formula of the stability coefficient w of the real scene image is: Where x n+Indicates the pixel value of the nth pixel in the positive sample set, x n- Represents the pixel value of the nth pixel in the negative sample set, N represents the sample threshold value, Represents the mean pixel value of all pixels in the real scene image.

[0054] By using correlation analysis of pixel values, we can reduce the impact of factors such as illumination and noise on detection results and improve the robustness of the algorithm in complex environments.

[0055] In this embodiment of the present invention, S2 includes the following sub-steps:

[0056] S21. Generate a stability weight of the real scene image based on the stability coefficient;

[0057] S22. Generate a symbol value for each pixel according to the stability weight of the real scene image.

[0058] In this paper, a randomized approach is used to generate stable weights, introducing a degree of randomness to each real-world image. This allows the algorithm to produce diverse weight distributions when processing different images, better adapting to a variety of complex industrial scenarios. Different images may have different feature distributions, and randomized weights prevent the algorithm from falling into a fixed pattern, improving its ability to capture diverse image features.

[0059] When calculating the random symbol value for a pixel, the average pixel value of its four neighboring pixels is used. This incorporates the pixel's local neighborhood information and considers the pixel's context within the image. This neighborhood information provides more clues to image features, allowing the generated symbol value to better reflect the pixel's local characteristics, helping to more accurately identify obstacle areas.

[0060] In the embodiment of the present invention, in S21, the calculation formula of the stability weight W is: Where w represents the stability coefficient of the real image, and rand(·) represents the random function. This function generates a random number between w and -w.

[0061] In the embodiment of the present invention, in S22, the calculation formula of the random symbol value F of the pixel point is: ;W represents the stability weight of the pixel, x represents the pixel value of the pixel, It represents the mean pixel value of the four neighboring pixels around the pixel point, and sign represents the sign function.

[0062] If necessary, the pixel values ​​can be de-dimensionalized to make the calculation reasonable.

[0063] In the embodiment of the present invention, in S3, the pixel points with a sign value of -1 are regarded as the obstacle area in front of the industrial robot body.

[0064] In this invention, the stability of different image regions is quantified through the calculation of stability coefficients and stability weights, and pixels are classified and weighted. Pixels with a symbol value of -1 exhibit distinct characteristics distinct from the background, and identifying them as obstacle regions can highlight the characteristics of obstacles. By determining the obstacle region in front of the industrial robot based on the symbol values ​​of the pixels, the abstract stability coefficient and symbol value can be converted into specific obstacle region information, providing a clear and specific target area for the industrial robot's obstacle avoidance and path planning, effectively ensuring its safe operation.

[0065] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A method for monitoring the operation of an industrial robot, characterized in that: The following steps are involved: S1. Collect real-scene images in front of the industrial robot body and calculate the stability coefficient based on two sample sets of the real-scene images; S2. Calculate the symbol value for each pixel based on the stability coefficient of the real scene image; S3. Determine the obstacle area in front of the industrial robot according to the symbol value of each pixel point in the real scene image; The S2 includes the following sub-steps: S21. Generate a stability weight of the real scene image based on the stability coefficient; S22. Generate a symbol value for each pixel according to the stability weight of the real scene image; In the above S21, the calculation formula of the stability weight W is: W=rand(w,-w); wherein w represents the stability coefficient of the real scene image, and rand(·) represents the random function; In S22, the calculation formula of the random symbol value F of the pixel point is: Where W represents the stability weight of the pixel, x represents the pixel value of the pixel, represents the mean pixel value of the four neighboring pixels around the pixel point, and sgn(·) represents the sign function; The S1 includes the following sub-steps: S11, collecting real-scene images in front of the industrial robot body; S12, extracting a positive sample set and a negative sample set of the real scene image; S13. Calculate a stability coefficient based on the positive sample set and the negative sample set of the real scene image; In S12, the pixel value mean of all pixels in the real scene image is extracted, and the pixel values ​​greater than the pixel value mean are used as the positive sample set, and the pixel values ​​less than or equal to the pixel value mean are used as the negative sample set; The S13 includes the following sub-steps: S131, taking the minimum value between the number of samples in the positive sample set and the number of samples in the negative sample set as the sample threshold value N; S132, sorting the pixel values ​​of the positive sample set and the negative sample set from large to small, and extracting the top N pixel points in the positive sample set and the top N pixel points in the negative sample set; S133. Calculate the stability coefficient of the real scene image based on the top N pixels in the positive sample set and the top N pixels in the negative sample set; In the above S133, the calculation formula of the stability coefficient w of the real scene image is: Where x n+ Indicates the pixel value of the nth pixel in the positive sample set, x n- Represents the pixel value of the nth pixel in the negative sample set, N represents the sample threshold value, Represents the mean pixel value of all pixels in the real scene image.

2. The operation monitoring method of an industrial robot according to claim 1, characterized in that: In S3, the pixel points with a sign value of -1 are regarded as the obstacle area in front of the industrial robot body.

Images