Method for identifying a carpet by a robot, identification device and robot

Abstract

The application discloses a method for identifying a carpet by a robot, an identification device and the robot, and comprises the following steps: S1, a robot controls a light supplement lamp to work, so that the light supplement lamp projects a light projection on a detection area of a walking surface, and then an image with the light projection in the detection area is acquired through a camera; S2, the robot acquires the smoothness of the edge of the light projection in the image with the light projection through a smoothness evaluation function; S3, the robot acquires the height between the detection surface and the light supplement lamp according to the position of the light projection in the image with the light projection, the position of the light supplement lamp and the emission angle of the light supplement lamp; and S4, the robot compares the smoothness of the edge of the light projection and the height between the detection surface and the light supplement lamp with corresponding set values respectively to determine whether the walking surface is a carpet. The application cooperates the camera and the light supplement lamp to identify the carpet, so that the detection data is more accurate, and the detection result is more reliable.

Description

Technical Field

[0001] This invention relates to the field of information transmission technology, and specifically to a method, identification device, and robot for a robot to identify carpets. Background Technology

[0002] Robotic vacuum cleaners have gained widespread use. Essentially, they use various sensors to perceive the external world and control their behavior based on sensor feedback. However, some systems are limited by their built-in sensors and processing algorithms, making it unable to detect special items like carpets and cleaning carpets like regular floors, failing to effectively remove debris and dirt. Others rely on a simple method using a single sensor. For example, a single proximity sensor for material recognition uses the difference in light reflectivity between hard surfaces and carpets, but it has limitations in distinguishing between rough, hard surfaces like black or gray, and short-pile carpets like white ones. Summary of the Invention

[0003] To address the aforementioned problems, this invention discloses a method, device, and robot for carpet identification. This application utilizes a camera and supplementary lighting to identify carpets by detecting the smoothness of the current walking surface and the detection height, thereby improving the accuracy of the detection results and enabling the robot to better plan its cleaning process. The specific technical solution is as follows:

[0004] A method for robot carpet recognition includes the following steps: S1: The robot controls a supplementary light to project a light projection onto a detection area of ​​the walking surface, and then acquires an image of the light projection in the detection area using a camera; S2: The robot obtains the smoothness of the edges of the light projection in the image using a smoothness evaluation function; S3: The robot obtains the height between the detection surface and the supplementary light based on the position of the light projection in the image, the position of the supplementary light, and the emission angle of the supplementary light; S4: The robot determines whether the walking surface is a carpet by comparing the smoothness of the light projection edge and the height between the detection surface and the supplementary light with corresponding set values. Compared with existing technologies, this application acquires an image with a supplementary light projection using a camera, and then obtains the smoothness of the current detection surface and the detection height between the detection surface and the supplementary light based on the pixels and position of the light projection in the image and the set angle of the supplementary light. Carpet recognition is performed through the cooperation of the camera and the supplementary light, resulting in more accurate detection data and more reliable detection results.

[0005] Furthermore, in step S1, the supplementary light projects a light projection pattern of a predetermined shape onto the detection area, and a calculation mark is provided on the side of the light projection pattern of the predetermined shape. The calculation mark makes robot calculation more convenient.

[0006] Further, in step S2, the robot obtains the smoothness of the edges of the light projection in the image from the image with light projection through the smoothness evaluation function, including the following steps: S21: The robot identifies the edges of the light projection in the image with light projection and obtains the edge lines; S22: The number of pixels on the edge lines is obtained, and then the vertical position value of each pixel on the edge lines is obtained; S23: The vertical position values ​​of adjacent pixels on the edge lines are subtracted, and then the average value of the absolute values ​​of the differences is obtained. This average value is the smoothness of the edges of the light projection in the image with light projection.

[0007] Furthermore, in step S21, the robot identifies the edge of the light projection as a straight line by using the brightness threshold of the light projection in the image with the light projection.

[0008] Further, in step S3, the robot obtains the height between the detection surface and the supplementary light using the following steps: S31: The robot pre-obtains the preset angle between the center line of the supplementary light and the center line of the camera; S32: The robot projects a light projection of a set pattern onto the detection area of ​​the walking surface using the supplementary light. The side of the light projection of the set pattern has a calculation mark, and the robot obtains an image of the light projection of the set pattern through the camera; S33: The robot obtains a vertical line in the image where the calculation mark of the light projection of the set pattern is located and which is perpendicular to the emission direction of the supplementary light, and then obtains the vertical distance between the vertical line and the vertical projection point of the supplementary light on the image; S34: The robot uses the preset angle and the vertical distance to obtain the height between the detection surface and the supplementary light using the arctangent function.

[0009] Furthermore, the light projection of the set graphic is a convex shape after being rotated 90 degrees, the calculated identifier is a rectangular protrusion of the convex shape after being rotated 90 degrees, and the vertical distance is the vertical distance between the bottom or top line of the rectangular protrusion of the convex shape after being rotated 90 degrees and the vertical projection point of the fill light on the image.

[0010] Furthermore, in step S4, if the smoothness of the edge of the light projection is less than the smoothness setting value, and the height between the detection surface and the supplementary light is less than the height setting value, then the robot determines that the walking surface is a carpet.

[0011] Furthermore, the height setting value is the height between the normal ground and the supplementary light.

[0012] A carpet recognition device includes a camera and a supplementary light. The camera is positioned perpendicular to the ground and facing downwards, and the centerline of the supplementary light forms a preset angle with the centerline of the camera. This carpet recognition device has a simple structure, high recognition accuracy, and reduces the production cost of robots.

[0013] Furthermore, the preset angle is any angle between 20 degrees and 70 degrees.

[0014] Furthermore, the front end of the camera is provided with a projection sheet, and the projection sheet is provided with a through hole with a set pattern.

[0015] A robot, wherein the aforementioned carpet recognition device is provided. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the method for robot carpet recognition in one embodiment of the present invention.

[0017] Figure 2 This is a projection pattern of the light projection of the supplementary light lamp described in one embodiment of the present invention;

[0018] Figure 3 This is a schematic diagram of the structure of the carpet recognition device according to one embodiment of the present invention;

[0019] Figure 4 This is a partial image with light projection as described in one embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0021] like Figure 1 As shown, a method for robot carpet recognition includes the following steps: S1: The robot controls a supplementary light to project a light projection onto the detection area of ​​the walking surface, and then acquires an image of the light projection in the detection area through a camera; S2: The robot obtains the smoothness of the edge of the light projection in the image using a smoothness evaluation function; S3: The robot obtains the height between the detection surface and the supplementary light based on the position of the light projection in the image, the position of the supplementary light, and the emission angle of the supplementary light; S4: The robot determines whether the walking surface is a carpet by comparing the smoothness of the edge of the light projection and the height between the detection surface and the supplementary light with corresponding set values. Compared with existing technologies, this application acquires an image with a supplementary light projection through a camera, and then obtains the smoothness of the current detection surface and the detection height between the detection surface and the supplementary light based on the pixels and position of the light projection in the image and the set angle of the supplementary light. Carpet recognition is performed through the cooperation of the camera and the supplementary light, resulting in more accurate detection data and more reliable detection results.

[0022] In one embodiment, in step S1, the supplementary light projects a light projection pattern of a predetermined shape onto the detection area, and a calculation mark is provided on the side of the light projection pattern of the predetermined shape. The calculation mark makes robot calculation more convenient.

[0023] In one embodiment, step S2, where the robot obtains the smoothness of the edges of the light projection in the image using a smoothness evaluation function, includes the following steps: S21: The robot identifies the edges of the light projection in the image and identifies them as straight lines to obtain edge lines; S22: The robot obtains the number of pixels on the edge lines and then obtains the vertical position value of each pixel on the edge lines; S23: The robot calculates the difference between the vertical position values ​​of adjacent pixels on the edge lines and then obtains the average of the absolute values ​​of the differences. This average value is the smoothness of the edges of the light projection in the image. In step S21, the robot identifies the edges of the light projection as edge lines using a brightness threshold for the light projection in the image, and calculates the number of pixels on the edge lines by counting the corresponding pixels in the image acquired by the camera. The summation function is:

[0024] ;

[0025] Where A is the smoothness of the edge of the light projection, n is the number of pixels on the edge line, and x is the vertical position value of the pixel.

[0026] like Figure 4 As shown, the vertical position values ​​are the pixel positions in the vertical direction of the image, determined by the robot's line recognition (i.e., the data on the right side of the image). The vertical position values ​​of the edge lines in the image are 7, 6, 5, 6, and 6. The smoothness of the edge lines in the image is:

[0027] A=(|0-7|+|7-6|+|6-5|+|5-6|+|6-6|) / 5.

[0028] In one embodiment, step S3, where the robot obtains the height between the detection surface and the supplementary light, includes the following steps: S31: The robot pre-obtains a preset angle between the centerline of the supplementary light and the centerline of the camera; S32: The robot projects a light projection of a set pattern onto the detection area of ​​the walking surface using the supplementary light. The side of the light projection of the set pattern has a calculation mark, and the robot obtains an image of the light projection of the set pattern through the camera; S33: The robot obtains a vertical line in the image where the calculation mark of the light projection of the set pattern is located and is perpendicular to the emission direction of the supplementary light, and then obtains the vertical distance between the vertical line and the vertical projection point of the supplementary light on the image; S34: The robot uses the preset angle and the vertical distance to obtain the height between the detection surface and the supplementary light using the arctangent function, i.e., H=W*atan(θ), where H is the height between the detection surface and the supplementary light, W is the vertical distance, and θ is the preset angle.

[0029] like Figure 2 As shown, the light projection of the set graphic is a convex shape rotated 90 degrees, the calculation mark is a rectangular protrusion of the convex shape rotated 90 degrees, and the vertical distance is the vertical distance between the bottom or top of the rectangular protrusion 103 (i.e., the calculation mark) of the convex shape rotated 90 degrees and the vertical projection point of the fill light on the image. The shape of the light projection of the set graphic can be a circle, ellipse, triangle, etc., and then the calculation mark is added on the basis of these graphics. The height can also be determined by the distance of one of the sides, without adding the side protrusions (i.e., the calculation mark). Adding the side protrusions is mainly to avoid insufficient illumination range when the height changes.

[0030] In one embodiment, in step S4, if the smoothness of the edge of the light projection is less than a smoothness setting value, and the height between the detection surface and the supplementary light is less than a height setting value, then the robot determines that the walking surface is a carpet. The height setting value is the height between the normal ground and the supplementary light. The edge of the light projected onto the ground is very straight. When encountering a bright surface, this edge can maintain a straight line with high smoothness. When projected onto a carpet, due to the influence of the carpet fibers, the edge of the light is not smooth, and the smoothness is relatively low. When the robot is on a carpet, because the wheels will sink into the carpet, the height between the detection surface and the supplementary light will be less than the height between the normal ground (hard ground) and the supplementary light. The height between the normal ground (hard ground) and the supplementary light can be detected when the robot is working, or it can be the installation height detected by other tools after the supplementary light is installed. By judging the straightness smoothness and the height H, it can be determined whether the ground material is a carpet.

[0031] like Figure 3As shown, a carpet recognition device performs the aforementioned robot carpet recognition method. The carpet recognition device includes a camera 101 and a supplementary light 102. The camera 101 is perpendicular to the ground and faces downwards. The centerline of the supplementary light 102 forms a preset angle 106 with the centerline of the camera 101. This carpet recognition device has a simple structure, high recognition accuracy, and reduces the production cost of the robot. The preset angle 106 can be any angle between 20 and 70 degrees. If the camera 101 and the supplementary light 102 are located at the bottom of the robot, the angle can be selected from 30, 45, or 60 degrees. The preset angle 106 can also be outside the range of 20 to 70 degrees. Values ​​outside this range may prevent the camera 101 from acquiring valid data. If the camera 101 can acquire valid data, the preset angle can be outside the range of 20 to 70 degrees. The distance between camera 101 and fill light 102 is set according to actual conditions. When camera 101 and fill light 102 are set close to each other, camera 101 should not block the light projection of fill light 102. When camera 101 and fill light 102 are set far apart, the vertical projection point of fill light 102 on the ground should not exceed the image acquisition range of camera 101. The setting of the distance between camera 101 and fill light 102 affects the selection of preset angle 106. When camera 101 and fill light 102 are set close to each other, preset angle 106 is selected between 20 degrees and 45 degrees. When camera 101 and fill light 102 are set far apart, preset angle 106 is selected between 45 degrees and 70 degrees. That is, the closer fill light 102 and camera 101 are, the smaller the preset angle 106 is, and the farther the fill light 102 and camera 101 are, the larger the preset angle 106 is. The front end of the camera 101 is provided with a projection sheet, and the projection sheet is provided with a through hole with a set pattern. The shape of the through hole is set according to the required projection pattern and is not limited. Generally, a rectangle that is easy to calculate is used. To prevent insufficient illumination range when the height changes, small rectangular protrusions can be set on the side of the rectangular through hole.

[0032] A robot is provided, incorporating the aforementioned carpet recognition device. This device can be positioned at the bottom, front, rear, or side of the robot, as long as the camera of the carpet recognition device is perpendicular to the ground and facing downwards. The robot can calculate the image displacements dx and dy by matching and aligning corner points. The actual robot displacements X and Y are calculated by multiplying the image displacements dx and dy by a coefficient related to height. The actual robot displacements X and Y are calculated using a pinhole camera model: dx = f*X / H, dy = f*Y / H; This yields X = dx*H / f, Y = dy*H / f, where f is the focal length of the camera and H is the height detected by the robot.

[0033] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0034] The above embodiments are merely illustrative of several implementations of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A method for a robot to identify carpets, characterized in that, The method includes the following steps: S1: The robot controls the auxiliary light to project a light projection onto the detection area of ​​the walking surface, and then uses a camera to acquire an image of the light projection in the detection area. S2: The robot obtains the smoothness of the edges of the light projection in the image from the image with light projection through a smoothness evaluation function; S3: The robot obtains the height between the detection surface and the supplementary light based on the position of the light projection in the image with light projection, the position of the supplementary light, and the emission angle of the supplementary light; S4: The robot determines whether the walking surface is a carpet by comparing the smoothness of the edge of the light projection and the height between the detection surface and the supplementary light with the corresponding set values. In step S4, if the smoothness of the edge of the light projection is less than the smoothness setting value, and the height between the detection surface and the supplementary light is less than the height setting value, then the robot determines that the walking surface is a carpet. In step S2, the robot obtains the smoothness of the edges of the light projection in the image from the image with light projection through a smoothness evaluation function, including the following steps: S21: The robot identifies the edges of light projections in an image with light projections and obtains the straight lines of the edges; S22: Get the number of pixels on the edge line, and then get the vertical position value of each pixel on the edge line; S23: Difference the vertical position values ​​of adjacent pixels on the edge line, and then obtain the average of the absolute values ​​of the differences. This average value is the smoothness of the edge of the light projection in the image with light projection.

2. The method for robot identification of carpets according to claim 1, characterized in that, In step S1, the supplementary light projects a light projection pattern of a set shape onto the detection area, and the side of the light projection pattern of the set shape is provided with a calculation mark.

3. The method for robot identification of carpets according to claim 1, characterized in that, In step S21, the robot identifies the edge of the light projection as a straight line by using the brightness threshold of the light projection in the image with light projection.

4. The method for robot identification of carpets according to claim 1, characterized in that, In step S3, the robot obtains the height between the detection surface and the supplementary light, including the following steps: S31: The robot pre-obtains the preset angle between the center line of the supplementary light and the center line of the camera; S32: The robot projects a light projection of a set pattern onto the detection area of ​​the walking surface using a supplementary light, and acquires an image of the light projection of the set pattern through a camera; S33: Obtain the vertical line that is perpendicular to the emission direction of the fill light and the calculation mark of the light projection of the set graphic in the image. Then obtain the vertical distance between the vertical line and the vertical projection point of the fill light on the image. S34: The robot uses the arctangent function to obtain the height between the detection surface and the supplementary light by using a preset angle and vertical distance.

5. The method for robot identification of carpets according to claim 4, characterized in that, The light projection of the set graphic is a convex shape after being rotated 90 degrees, the calculation mark is a rectangular protrusion of the convex shape after being rotated 90 degrees, and the vertical distance is the vertical distance between the bottom or top of the rectangular protrusion of the convex shape after being rotated 90 degrees and the vertical projection point of the fill light on the image.

6. The method for robot identification of carpets according to claim 1, characterized in that, The height setting is the height between the ground and the fill light.

7. A carpet identification device, characterized in that, The method for robot carpet recognition as described in any one of claims 1 to 6, wherein the carpet recognition device includes a camera and a supplementary light, the camera being perpendicular to the ground and facing downwards, and the center line of the supplementary light being at a preset angle to the center line of the camera; The preset angle is any angle between 20 degrees and 70 degrees; The camera has a projection sheet at its front end, and the projection sheet has a through hole with a set pattern.

Images