Navigation mapping method for constructing machine surrounding image
By acquiring and processing real-time images of the target area and establishing a 2D image coordinate system, the problem of difficulty in navigating robots in environments such as swimming pools and lawns is solved, and a broad field of view and high-precision navigation effect is achieved.
Patent Information
- Application Number
- PCT/CN2023/108771
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-08
- Filing Date
- 2023-07-23
- Publication Date
- 2024-11-14
AI Technical Summary
It is difficult for existing robots to obtain sufficient feature points for navigation through traditional sensors in environments such as swimming pools and lawns, which makes it difficult to achieve navigation methods to avoid obstacles, especially due to the limitation of sensor height and scarcity of feature points.
By obtaining the real-time image of the target area, performing binarization processing, establishing a 2D image coordinate system, determining the real-time position of the machine in the target area, and using a multi-view image fusion algorithm to obtain a top-view 2D image to intuitively display the positions of the machine and obstacles. Expand the field of view to avoid the limitations caused by the sensor setting on the machine.
It realizes the acquisition of a wide field of view of the robot outside the robot, intuitively determines the machine position and obstacle position, especially at the boundary, and improves the feasibility and accuracy of navigation.
Smart Images

Figure CN2023108771_14112024_PF_FP_ABST
Abstract
Description
Mapping method for constructing machine external images for navigation Technical Field
[0001] The present invention relates to the field of navigation mapping technology, and in particular to a navigation mapping method for constructing an external image of a machine. Background Art
[0002] Currently, all robots on the market have the function of photo recognition and can avoid obstacles in a timely manner. For example, the sensors of water cleaning robots in swimming pools or lawn mowing robots are all set on the machines. However, in actual use, since there are fewer feature points on swimming pools and lawns, and the height of the sensor is affected by the height of the robot, the range of photos that can be obtained is very small and the number of feature points obtained is also small. Therefore, the pictures obtained by setting the sensor in this way will make it difficult for the machine to identify and locate them, which makes the navigation method of avoiding obstacles difficult to implement in actual applications.
[0003] Although both positioning and radar can effectively avoid obstacles, both the Beidou positioning system and radar judgment have certain limitations, especially in water surface cleaning robots or lawn mowing robots. This is because these robots have a target area in which they work and they only operate within that area, and the boundary of that area cannot be detected by radar. Positioning can only determine the position of a point, and the image of the robot's exterior cannot be intuitively displayed. Therefore, both have certain disadvantages.
[0004] Summary of the Invention
[0005] The object of the present invention is to provide a navigation mapping method for constructing an external image of a machine, so as to solve the problems raised in the above-mentioned background technology.
[0006] To achieve the above-mentioned purpose, a navigation mapping method for constructing an image of the exterior of a machine is provided, which includes the following method steps:
[0007] Acquire real-time images of the target area;
[0008] Processing the real-time image to obtain a 2D image of the target area;
[0009] Based on the coverage area of the machine in the 2D image, the real-time position of the machine in the target area is obtained.
[0010] As a further improvement of the present technical solution, the acquisition of the real-time image and real-time position includes the following steps:
[0011] S1. At time T, take a photo of the target area in a manner that covers the target area to obtain a real-time image of the outside of the machine at time T;
[0012] S2, processing the real-time image at time T to obtain a 2D image;
[0013] S3, identifying the coverage area of the machine in the 2D image at time T, and determining the position of the machine in the 2D image at time T based on the coverage area;
[0014] S4. After time t, T=T+t is obtained, and steps S1-S4 are repeated.
[0015] As a further improvement of the present technical solution, the processing of the real-time image in S2 includes a binarization operation, and the operation steps are as follows:
[0016] Get the grayscale threshold;
[0017] Based on the grayscale threshold, each pixel in the real-time image is judged:
[0018] If the grayscale value of the pixel is greater than the grayscale threshold, the grayscale value of the pixel is set to 1;
[0019] If the grayscale value of the pixel is less than the grayscale threshold, the grayscale value of the pixel is set to 0.
[0020] As a further improvement of the present technical solution, in the 2D image, the area with the pixel value of "0" is determined as the area where the machine can walk, and the area with the pixel value of "1" is determined as the area where the machine cannot walk.
[0021] As a further improvement of the present technical solution, in S4, the position of the machine in the 2D image obtained each time is recorded and a coordinate system is established;
[0022] include:
[0023] Setting a coordinate origin in the 2D image and establishing a coordinate system based on the coordinate origin;
[0024] In the coordinate system, the coordinates of specific points on the machine coverage surface are recorded.
[0025] As a further improvement of the present technical solution, the speed calculation formula of the machine in the coordinate system is as follows: P(x,y);
[0026] Where P(x,y) is the coordinate of a specific point recorded in the coordinate system; v x is the velocity on the x-axis at a specific point t; v y is the velocity on the y-axis at a specific point t; x T+t is the horizontal coordinate of a specific point after time t; x T is the horizontal coordinate of a specific point before time t; y T+tis the ordinate of a specific point after time t; y T is the vertical coordinate of a specific point before time t.
[0027] As a further improvement of the present technical solution, the photographing of the target area in S1 is completed by using a camera, and one or more cameras are provided so that the photographing area obtained by photographing covers the target area.
[0028] As a further improvement of the present technical solution, at least two cameras are used to obtain the real-time image, and the cameras are set up at a high point in the target area;
[0029] The photos taken by the multiple cameras are highly consistent;
[0030] The plurality of cameras take photos to form a photo area, and each photo area has at least one other photo area forming an intersection area with it;
[0031] The photographed areas are fused with the help of the intersection area to obtain a 2D image.
[0032] As a further improvement of the present technical solution, at least three cameras are used to obtain the real-time image, and at least one of the cameras is selected as a high-mounted camera;
[0033] The high-position photographing area formed by the high-position camera photographing and other photographing areas all form an intersection area, and the combination of other photographing areas can cover the entire target area;
[0034] The photographed areas are fused with the help of the intersection area to obtain a 2D image.
[0035] As a further improvement of this technical solution, the fusion of the photographing area adopts a multi-view image fusion algorithm, and its algorithm formula is as follows:
[0036] Where x i and x j are the horizontal coordinates of the feature points of the two photographic areas in the intersection area; y i and y j are the horizontal and vertical coordinates of the feature points of the two photographic areas in the intersection area; F is the basic matrix from one photographic area to another.
[0037] Compared with the prior art, the present invention has the following beneficial effects:
[0038] In this navigation mapping method for constructing an image of the machine's exterior, a 2D image of the machine's exterior can be obtained from a bird's-eye view. If the machine is in the target area, the 2D image obtained by taking the photo will include a bird's-eye view of the machine. This bird's-eye view can be used to determine the machine's location. Both the machine and obstacles in the target area can be intuitively displayed in the 2D image.
[0039] More importantly, the machine changes its position in a 2D image, making it more intuitive to determine its position, especially at the boundary;
[0040] Moreover, since the sensor is not set on the machine, the obtained 2D image is not limited by the machine itself, and the field of view of the obtained 2D image becomes wider. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] FIG1 is a schematic flow chart of the overall method steps of the present invention;
[0042] FIG2 is a schematic flow chart of the binarization operation steps of the present invention;
[0043] FIG3 is a schematic diagram of a photographing area under a camera of the present invention;
[0044] FIG4 is a schematic diagram of the photographing area under two cameras of the present invention;
[0045] FIG5 is a schematic diagram showing the principle of fusion of two photographing areas according to the present invention;
[0046] FIG6 is a schematic diagram of a high-mounted photographing area of a high-mounted camera according to the present invention;
[0047] FIG7 is a schematic diagram of the target area determination principle of the present invention;
[0048] FIG8 is a schematic diagram of the machine movement principle of the present invention. DETAILED DESCRIPTION
[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0050] Currently, all robots on the market have the function of photo recognition and can avoid obstacles in a timely manner. For example, the sensors of swimming pool cleaning robots or lawn mowing robots are all set on the machines. The features of swimming pools and lawns are very similar. The sensors are affected by the height of the robots, and the field of view of the photos they can take is very small. Therefore, the pictures obtained in this way will make it difficult for the machine to recognize, which makes the navigation method of avoiding obstacles difficult to implement in actual applications.
[0051] To this end, the present invention provides a navigation mapping method for constructing an external image of a machine, which includes the following method steps:
[0052] Acquire real-time images of the target area;
[0053] Process the real-time image to obtain a 2D image of the target area;
[0054] Get the real-time location of the machine in the target area based on the machine's coverage in the 2D image.
[0055] As shown in Figure 1, the specific steps of the mapping method are as follows:
[0056] First, a device for taking pictures is set up directly above the target area. The device can be a camera mounted on a drone, or a camera can be set up above the target area. Using a drone is more convenient, but the stability is not high (it is easily shaken by weather, resulting in blurred images), so it is suitable for indoor use, such as in a swimming pool; directly setting up a camera is more cumbersome, inflexible, and greatly affected by space because it is necessary to build a bracket to connect the camera, but it is more stable, so it is suitable for outdoor environments, such as on a lawn.
[0057] Next, S1 is performed. At time T, a photo is taken directly facing the target area in a manner covering the target area, and a real-time image is obtained by taking the photo. Specifically, the camera is used to implement the photo-taking operation to obtain a 2D image of the outside of the machine (i.e., a robot or other self-moving machine that can be controlled) at time T. If only one camera is set, the restriction of time T only determines the time of taking the photo. However, if multiple cameras are set, the restriction of time T can not only determine the time of taking the photo, but also ensure that the 2D images taken by the cameras are taken at the same time, so as to achieve the purpose of controlling variables and avoid differences in certain features of 2D images obtained by different cameras due to different photo-taking times.
[0058] It should be noted that, because a 2D image on the ground is to be obtained, the camera is required to take pictures facing the target area, so that a 2D image of the objects in the target area can be obtained from a bird's-eye view.
[0059] In addition, there is no limit on the number of cameras, which is determined based on the photography requirements and actual usage. However, it is necessary to ensure that the 2D image or combination of 2D images obtained by the camera can cover the entire target area. Generally, the larger the target area, the more cameras are set.
[0060] S2. performing a binarization operation on the 2D image at time T to convert the colored 2D image into a 2D image with only "0" and "1" pixel values;
[0061] As shown in Figure 2, the steps for binarization of a 2D image are as follows:
[0062] S2.1.1. Load a 2D image outside the machine;
[0063] S2.1.2. Converting a colored 2D image into a 2D image with only “0” and “1” pixel values;
[0064] When the RGB value is less than the grayscale threshold (0, 240, 255), it is converted to 1; when it is greater than the grayscale threshold (0, 240, 255), it is converted to 0.
[0065] S3. Identify the machine coverage in the 2D image at time T, and determine the position of the machine in the 2D image at time T based on the coverage. That is, the point determining the position has an area attribute, and the area attribute is determined based on the machine coverage.
[0066] Since it is necessary to track the machine, the above steps must be repeated to obtain the changes in the machine's position. Therefore, after S4 and time t, T=T+t is obtained, and steps S1-S4 are repeated, where: t is a shooting time interval of the camera. In one embodiment, the shooting frequency is preferably controlled at 5 frames per second, that is, t=0.2s.
[0067] Through the above steps, a 2D image of the outside of the machine can be obtained from a bird's-eye view. If the machine is in the target area, the 2D image obtained by taking the photo will include a bird's-eye view of the robot. This bird's-eye view can be used to determine the machine's current position in the lawn or swimming pool. The 2D image obtained after acquisition and processing can intuitively display the machine and obstacles in the target area.
[0068] More importantly, the robot changes its position in the 2D image as a two-dimensional image, which makes the determination of its position more intuitive, especially at the boundaries of the 2D image.
[0069] Moreover, since the image sensor is not installed on the machine when acquiring the image, the obtained 2D image is not limited by the height of the machine itself, and the obtained 2D image has a wider field of view. In addition, the acquisition of 2D images is not limited to the above-mentioned form. Other forms are described in the following embodiments:
[0070] In the first embodiment, a camera is provided in this embodiment, and the target area is directly photographed by the camera, as shown in FIG3 . In the figure, A is the photographing area photographed by the camera. It can be seen that the photographing area A can cover the entire target area a. The advantage of this embodiment is that there is no need to consider the influence of variables, because one photographing area A can cover the entire target area a, and there is no need to fuse the photographing areas.
[0071] In the above embodiment, when the target area a is small, one camera can be used to take pictures. Usually, a camera with a medium viewing angle is used. When the shooting height is below 50 meters, the area of the lake that can be photographed can cover about 500 square meters. When the shooting height is below 10 meters, the area of the lake that can be photographed can cover about 100 square meters. However, considering that the position of the machine also needs to be determined, this leads to a great limitation on the scenes that can be used for taking pictures with one camera.
[0072] In the second embodiment, at least two cameras are set to take pictures of the target area so that the photographed area can still cover the target area when the shooting height is lowered. At this time, it is necessary to fuse multiple photographed areas; therefore, in the obtained photographed areas, each photographed area has at least one other photographed area forming an intersection area with it; the photographed areas are then fused with the help of the intersection area.
[0073] Specifically, feature points are found in the intersection area, and then the multi-view image fusion algorithm uses the common feature points of the two shooting areas in the intersection area to fuse the two. The algorithm formula is as follows:
[0074] Where x i and x j are the horizontal coordinates of the feature points of the two photographic areas in the intersection area; y i and y j are the horizontal and vertical coordinates of the feature points of the two photographic areas in the intersection area; F is the basic matrix from one photographic area to another.
[0075] Assume that two cameras are set up and the camera shooting heights are consistent, as shown in Figure 4. Because the heights are consistent, the same type of cameras can obtain the same shooting areas. The two cameras obtain shooting areas B and C. If a single shooting area (i.e., shooting area B or shooting area C) cannot cover the target area b, but the combination of the two can cover the target area b, feature points are selected in the intersection area (i.e., the shadow part) of shooting area B and shooting area C. The more prominent the selected feature points are in the intersection area, the better, so that it is easier to determine the coordinates of the feature points during fusion.
[0076] Assume that there are two feature point coordinates, point 1 and point 2, and obtain the coordinates of point 1 (x1, y1) and point 2 (x2, y2). Then, the photographing area B is fused into the photographing area C. As shown in Figure 5, point 1′ and point 1″ are fused, and point 2′ and point 2″ are fused. Then, the target area b′ and target area b″ are fused. After fusion, the complete target area b can be obtained.
[0077] Preferably, when taking pictures, the machine is placed in the intersection area, so that there are machines in the obtained photographing area. Then, the machine is used as the basic feature point, and other feature points are selected around it. In this way, the machine itself is a significant feature. Then, with the help of the feature points around it, a feature point group is obtained. Against the background of the machine, the feature points around it also have a certain significance, so that the fused target area is more accurate.
[0078] Moreover, during fusion, the basic feature points can be used as the targets of the displacement parameters, and the remaining feature points can be used as the targets of the rotation parameters, so as to ensure that the photographed area is accurately positioned during fusion. Secondly, multiple feature points are used to ensure the accuracy of rotation. Even if a few feature points deviate, they will be corrected by other feature points.
[0079] However, if there is only one machine, then only two cameras can be set up, because a machine has only one position in the 2D image captured at the same time; if there are two machines, then three cameras can be set up; and so on, the number of machines is D, and the number of cameras is D+1.
[0080] It should be noted that when there are more than two cameras, the photo areas can be merged in pairs according to the above principle, which will not be elaborated here.
[0081] In a third embodiment, when the number of cameras is three or more, at least one camera is selected as a high-position camera;
[0082] The high-mounted camera forms a photographing area that intersects with other photographing areas. That is, the high-mounted camera takes photos at a higher height than other cameras, while the other cameras are at the same height. The combination of other photographing areas can cover the entire target area. In this way, a large photographing area can be obtained through the high-mounted camera.
[0083] As shown in FIG6 , for the same feature point, there may be certain differences in perspective due to changes in the photographing position, and usually in order to save the number of cameras, the edges of the photographing area are fused. The closer to the edge, the more obvious the difference in perspective will be, which leads to deviations during fusion. In this embodiment, a high-mounted camera is set to determine the feature point at the intersection of the photographing area A of the high-mounted camera and the intersection area, so that the feature point falls in the middle of the photographing area A. Then, the feature points in the photographing areas B and C are corrected through the high-mounted photographing area A to reduce the impact of the perspective deviation.
[0084] Moreover, the final fusion is performed on the photographed areas B and C, so it will not affect the presentation of the target area after fusion.
[0085] In addition, there is another implementation method. Since the high-mounted camera will have cameras covered in the high-mounted photographing area, this feature is used as verification to further improve the accuracy of the target area after fusion. Especially when multiple high-mounted cameras are set up, the camera is used as a feature to assist in the fusion of the photographing area covered by the high-mounted photographing area and the photographing area covered by other high-mounted photographing areas.
[0086] Based on the above embodiment, the following application scenario is set:
[0087] First, as shown in FIG7 , the area with a pixel value of “0” is determined as a machine-walkable area, and the area with a pixel value of “1” is determined as a machine-inaccessible area. Referring to FIG8 , the blank area is the machine-walkable area, and a boundary is formed at the edge of the blank area, that is, the dividing line formed by the outermost “0” pixel block and the “1” pixel block. Suppose that floating objects are salvaged from the surface of a swimming pool. At this time, the RGB pixels of the light blue water are (0, 240, 255). Then, the original RGB image (i.e., color image) is converted into a pixel grayscale image with only “0” and “1”. When the RGB value is less than (0, 240, 255), it is converted into “1”, and when it is greater than (0, 240, 255), it is converted into “0”.
[0088] According to the characteristics of grayscale images, whose pixel values are only "0" and "1", find the largest "0" connected target area range, find the left, right, top and bottom boundaries respectively, and define this area as the walking range of the machine.
[0089] Record the position of the machine in the 2D image each time and establish a coordinate system;
[0090] Recording with the coordinates of specific points on the machine coverage area;
[0091] The length of a straight line passing through a specific point perpendicular to the direction of movement of the machine on the coverage surface is known, which makes it easy to calculate the area swept by the machine and determine the external contour of the machine.
[0092] In addition, the speed of the machine in the coordinate system is calculated as follows: P(x,y);
[0093] Where P(x,y) is the coordinate of a specific point recorded in the coordinate system; v x is the velocity on the x-axis at a specific point t; v y is the velocity on the y-axis at a specific point t; x T+t is the horizontal coordinate of a specific point after time t; x T is the horizontal coordinate of a specific point before time t; y T+t is the ordinate of a specific point after time t; y T is the vertical coordinate of a specific point before time t.
[0094] Identify the machine in the target area. When the color of the machine is black, take a picture and get the RGB pixel value (128, 128, 128). Water is light blue or colorless, which is quite different from the gray RGB. Binarize the target area according to the RGB threshold (0, 240, 255) and find the coordinates of the robot's lower leftmost (X5, Y5), lower rightmost (X6, Y6), upper leftmost (X7, Y7), and upper rightmost (X8, Y8).
[0095] The camera is used to shoot at 5 frames per second to track the target. The tracking scheme selects two frames apart and records the position of one frame and the position of the second frame. The positions of the four corners of the first frame are the positions of the upper dotted box in Figure 8, the lower left coordinates of the upper dotted box are (X5, Y5), the lower right coordinates are (X6, Y6), the upper left coordinates are (X7, Y7), and the upper right coordinates are (X8, Y8). The positions of the four corners of the second frame are the lower dotted box, the lower left coordinates of the lower dotted box are (X9, Y9), the lower right coordinates are (X10, Y10), the upper left coordinates are (X11, Y11), and the upper right coordinates are (X12, Y12). Then the speed is calculated.
[0096] Finally, based on the robot's current speed, the direction of the robot in the two frames of image is used to determine whether the robot is heading towards the boundary. Then, the distance between the speed direction and the pixel point heading towards the boundary is calculated. When the robot is close to the boundary pixel point and the distance is 15, the robot is controlled or other operations are performed to avoid obstacles and navigate and plan the robot's path.
[0097] It should be noted that the figure is magnified for the convenience of intuitive observation of pixel blocks. In actual work, the more pixel blocks there are, the smoother the boundary of the target area will be.
[0098] Preferably, navigation lights are set along the outline of the machine to make the machine more conspicuous, and the angle of the light emitted by the navigation lights is not affected by the medium. That is to say, the navigation lights need to be set above the water surface so that the RGB value of the machine obtained by taking a photo is (128, 128, 128).
[0099] In addition, the machine changes its position in the 2D image as a two-dimensional image, which makes the determination of its position more intuitive. Especially at the boundary, the distance between the two can be accurately determined. The area passed by the machine can also be determined by the movement path to facilitate area calculation, making the application scenarios of this system more extensive, such as agricultural spraying, ground mapping, etc.
[0100] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.
Claims
1. Constructing a navigation map of the machine's external image, It is characterized in that The method comprises the following steps: Obtain real-time images of the target area; Processing the real-time image to obtain a 2D image of the target area; Based on the coverage of the machine in the 2D image, obtaining a real-time position of the machine in the target area; The acquisition of the real-time image and the real-time position includes the following steps: S1. At time T, take a photo of the target area in a manner of covering the target area to obtain a real-time image of the outside of the machine at time T; S2, processing the real-time image at time T to obtain a 2D image; S3, identifying the coverage area of the machine in the 2D image at time T, and determining the position of the machine in the 2D image at time T based on the coverage area; S4, after time t, T=T+t is obtained, and steps S1-S4 are repeated; The processing of the real-time image in S2 includes a binarization operation, and the operation steps are as follows: Get the grayscale threshold; Based on the grayscale threshold, each pixel in the real-time image is judged: If the grayscale value of the pixel is greater than the grayscale threshold, the grayscale value of the pixel is set to 1; If the grayscale value of the pixel is less than the grayscale threshold, the grayscale value of the pixel is set to 0; In the 2D image, the area where the gray value of the pixel point is "0" is determined as the machine-walkable area, and the area where the gray value of the pixel point is "1" is determined as the machine-unable area. Walking area; In said S4, the position of the machine in the 2D image obtained each time is recorded and a coordinate system is established; include: Setting a coordinate origin in the 2D image and establishing a coordinate system based on the coordinate origin; In the coordinate system, recording the coordinates of specific points on the coverage surface of the machine; The speed calculation formula of the machine in the coordinate system is as follows: P(x,y); Where P(x, y) is the coordinate of a specific point recorded in the coordinate system; v x is the velocity on the x-axis at a specific point at time t; v y is the velocity on the y-axis at a specific point at time t; x T+t is the horizontal coordinate of a specific point after time t; x T is the horizontal coordinate of a specific point before time t; y T+t is the ordinate of a specific point after time t; y T is the ordinate of a specific point before time t.
2. The navigation mapping method for constructing an external image of a machine according to claim 1, It is characterized in that The photographing of the target area in S1 is completed by using a camera, and one or more cameras are provided, and the one or more cameras are all provided outside the machine.
3. The navigation mapping method for constructing an external image of a machine according to claim 2, It is characterized in that At least two cameras are used to obtain the real-time image, and the cameras are set up at high points in the target area; The photographing heights of the multiple cameras are consistent; A plurality of cameras take photos to form a photo area, and each photo area has at least one other photo area to form an intersection area with it; The photographed areas are fused with the help of the intersection area to obtain a 2D image.
4. The navigation mapping method for constructing an external image of a machine according to claim 3, It is characterized in that At least three cameras are used to obtain the real-time image, and at least one of the cameras is selected as a high-mounted camera; The high-mounted photographing area formed by the high-mounted camera and other photographing areas all form an intersection area, and the combination of other photographing areas can cover the entire target area; The photographed areas are fused with the help of the intersection area to obtain a 2D image.
5. The navigation mapping method for constructing an external image of a machine according to claim 3 or 4, It is characterized in that The fusion of the photographing area adopts a multi-view image fusion algorithm, and its algorithm formula is as follows: In the formula, x i and x j are the horizontal coordinates of the feature points of the two photographing areas in the intersection area; y i and j are the horizontal and vertical coordinates of the feature points of the two photographing areas in the intersection area; F is the basic matrix from one photographing area to another photographing area.
Citation Information
Patent Citations
Automobile A-pillar perspective algorithm
CN110866863A
Robot control system and method based on visual boundary detection
CN113910225A
Statistical method and statistical device for cleaning area
CN114010104A
Path planning method, construction method and device, robot and storage medium
CN114527748A
Track measurement method and system of mobile robot based on multiple cameras
CN114549976A
