A high-precision docking method for laser vision fusion in complex lighting environments

Abstract

The present invention discloses a high-precision docking method for complex lighting environments using laser vision fusion, including an automatic docking system. The automatic docking system includes an active docking mechanism and a passive docking mechanism, wherein the active docking mechanism docks with the passive docking mechanism. The active docking mechanism includes a posture detection system, a control system, and a motion actuator. The posture detection system includes binocular cameras A and B, laser displacement sensors A and B, a servo electric cylinder, a force sensor A and B, and a vision module. The present invention utilizes a high-precision docking method based on the fusion of sensor data from computer vision, laser displacement sensors, force sensors, and other sensors. This method solves the problems of high equipment cost, inapplicability to mobile operations, and inapplicability to operations in complex lighting environments in traditional automatic docking systems. The system has low equipment cost, strong environmental adaptability, and excellent stability.

Description

Technical Field

[0001] The present invention relates to a high-precision docking method, in particular to a high-precision docking method for laser vision fusion in a complex lighting environment, belonging to the technical field of component docking. Background Art

[0002] Parts are the basic elements of a machine. A machine generally consists of one or more transmission components that receive external energy; an executive component that performs its production functions (such as the cutting tool in a machine tool); a transmission component that transmits the motion and power of the prime mover to the executive component (such as the gears and screw transmission mechanisms in machine tools); and a detection and control system that ensures the coordinated operation of all parts of the machine (such as the CNC system in machine tools). (That is, a machine consists of a prime mover, transmission component, executive component, and measurement and control component). Further decomposition of the machine reveals various parts.

[0003] Traditionally, the docking and assembly of large-scale six-degree-of-freedom components in space is usually done manually, which costs a lot of manpower and time, and has low efficiency. In recent years, researchers have developed automatic docking and assembly systems for large components in indoor environments. Laser trackers, iGPS, and computer vision are often used as posture detection equipment in automatic docking. The former two are expensive and cannot be applied to open and mobile outdoor working environments. Existing automatic docking systems based on computer vision have very high requirements for ambient lighting conditions and cannot work in complex lighting environments such as outdoors. Summary of the Invention

[0004] The purpose of the present invention is to provide a high-precision docking method for laser vision fusion in complex lighting environments, so as to solve the problem raised in the above background technology that the traditional large-scale spatial six-degree-of-freedom component docking and assembly usually rely on manual work, with high manpower and time costs and low efficiency of docking and assembly.

[0005] To achieve the above object, the present invention provides the following technical solution: comprising an automatic docking system, the automatic docking system comprising an active docking mechanism and a passive docking mechanism, the active docking mechanism docking with the passive docking mechanism;

[0006] The active docking mechanism includes a posture detection system, a control system and a motion actuator. The posture detection system consists of a binocular camera A, a binocular camera B, a laser displacement sensor A, a laser displacement sensor B, a servo electric cylinder, a force sensor A, a force sensor B and a vision module. The motion actuator consists of a parallel mechanism fixed platform and a parallel mechanism dynamic platform. The passive docking mechanism consists of an automatic locking mechanism. The parallel mechanism fixed platform is fixedly installed on the mobile platform installation surface. Force sensor A and force sensor B are respectively installed on both sides of the parallel mechanism dynamic platform. The force sensor A and force sensor B support the active docking body. The active docking body is aligned with the automatic locking mechanism. Columns A and column B are respectively fixedly installed on both sides of the parallel mechanism dynamic platform. The top of the column A is installed with a binocular camera A, and the top of the column B is installed with a binocular camera B, a laser displacement sensor A, and a laser displacement sensor B. The laser displacement sensor A is fixed to the column B, and one side of the laser displacement sensor B is connected to the output end of the servo electric cylinder.

[0007] As a preferred technical solution of the present invention, the high-precision docking method includes the following steps:

[0008] S1: Send the active docking mechanism to the docking implementation area and start the control system;

[0009] S2: Binocular camera B and laser displacement sensor A start working. Binocular camera B acquires images, performs orientation detection and feature extraction, and laser displacement sensor A acquires depth information.

[0010] S3: The visual module of the posture detection system accurately extracts the features of the passive docking mechanism within the field of view of the binocular camera B and integrates them with the distance information obtained by the laser displacement sensor A. It calculates the relative posture information of the passive docking mechanism and the active docking mechanism, and converts it into the motion control variable of the parallel mechanism dynamic platform through the Jacobian matrix. The parallel mechanism dynamic platform drives the active docking mechanism to approach the passive docking mechanism.

[0011] S4: When the active docking mechanism and the passive docking mechanism have reached the alignment state, the servo electric cylinder pushes the laser displacement sensor B above the column B to detect and calculate the position deviation between the two. After the servo electric cylinder retracts the laser displacement sensor B, the control system drives the parallel mechanism dynamic platform to drive the active docking mechanism to move and complete the docking. At the same time, the binocular camera A starts working. The information features of the binocular camera A (8), the laser displacement sensor A (10), the force sensor A (3), and the force sensor B (4) are extracted to determine whether there is an angle deviation between the active docking mechanism and the passive docking mechanism during the docking process, and the motion control amount of the parallel mechanism dynamic platform is corrected in time.

[0012] S5: The control system sends a remote control command to the automatic locking mechanism on the passive docking mechanism to complete the locking. After receiving the locking completion signal, the control system controls the parallel mechanism dynamic platform to be safely retracted and removed.

[0013] As a preferred technical solution of the present invention, the visual module encapsulates a visual algorithm and a feature extraction algorithm, and the visual algorithm is:

[0014] Step 1: Let the baseline length between the left and right fields of view of the binocular camera be b, the focal length be f, and the parallax of a feature point in the image between the left and right fields of view of the binocular camera be δ p , the depth value d of the feature point in the camera coordinate system can be easily obtained by the triangulation principle c for:

[0015]

[0016] Step 2: Take out three feature points obtained from the image, obtain their disparity respectively, calculate the depth information of the three points, convert the pixel coordinates to the camera coordinate system through the camera intrinsic parameters, and obtain the coordinates of the three points in the camera coordinate system, which are recorded as The extracted feature points are located on the plane of the object surface. The plane equation of the observed object surface can be obtained from the three points, which is recorded as:

[0017] aX+bY+cZ+d=0

[0018] in:

[0019]

[0020] c=(X2-X1)(Y3-Y1)-(X3-X1)(Y2-Y1)

[0021]

[0022] Step 3: According to the plane equation of the observed object surface in step 2, the installation position deviation between the laser displacement sensor and the midpoint of the binocular camera baseline in the camera coordinate system is δ=(δX cl ,δY cl ,δZ cl ), the coordinates of the surface point of the object measured by the laser displacement sensor in the camera coordinate system are p=(X p ,Y p ,Z p ), where X p =δX cl , Y p =δY cl , so point p=(δX cl ,δY cl ,Z p );

[0023] Step 4: Substitute point p into the plane equation:

[0024] aδX cl +bδY cl +cZ p +d=0

[0025] Get the depth of point p in the camera coordinate system:

[0026]

[0027] Step 5: Note that the depth measured by the laser displacement sensor at point p is d l , the same distance estimated by the binocular camera depth is d c ,in:

[0028] d c =Z p -δZ cl

[0029] Let the depth confidence of the feature points estimated by the binocular camera be c, then:

[0030]

[0031] As an optimal technical solution of the present invention, if the depth confidence c of the feature point is higher than the set threshold, the coordinates of the feature point are used to calculate the relative posture of the docking mechanism. If the depth confidence c of the feature point is lower than the set threshold, the data obtained this time is discarded.

[0032] As a preferred technical solution of the present invention, the feature extraction algorithm includes the following steps:

[0033] Step 1: Grayscale and distortion correct the acquired image as the original input;

[0034] Step 2: Based on the characteristics of the image in the field of view, the average pixel gradient of the full frame is obtained in the direction with less interference;

[0035] Step 3: Based on the pre-set window size of the maximum tilt angle of the docking mechanism, with the range of the rough boundary as the center, find the maximum gradient point of each row of pixels, perform straight line fitting on these points, and obtain the approximate boundary line equation;

[0036] Step 4: Divide the window into a new window according to the new line equation, and set the horizontal boundary according to the slope of the boundary to find the required sliding window height. Slide the sliding window along the extracted vertical boundary in this area, and find the upper and lower boundaries of the end row by row.

[0037] Step 5: Divide the window within the range, eliminate the interference of background objects, and find the maximum pixel gradient point by column. Fit the obtained points with a straight line to obtain the upper and lower boundaries;

[0038] Step 6: Calculate the intersection of the four obtained boundaries to obtain the four vertices at the end of the docking mechanism for the next pose calculation.

[0039] As a preferred technical solution of the present invention, the automatic locking mechanism includes a cross bar, a passive docking body, a lock column, a U-shaped connecting block, a slider and an electric slide rail. A passive docking body is fixedly installed on one side of the bottom end of the cross bar, a through hole is opened on one side of the passive docking body, and an electric slide rail is fixedly installed on the other side of the bottom end of the cross bar. The surface of the electric slide rail is slidably connected with a slider, and the bottom end of the slider is fixedly installed with a U-shaped connecting block, and a lock column is installed at the bottom of the U-shaped connecting block, and one end of the lock column is plugged into the through hole.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention provides a high-precision docking method for complex lighting environments using laser vision fusion. The high-precision docking method is based on the fusion of sensor data such as computer vision, laser displacement sensors, and force sensors. It solves the problems of high equipment cost, inapplicability to mobile operations, and inapplicability to operations in complex lighting environments in traditional automatic docking systems. The system has low equipment cost, strong environmental adaptability, high operating precision, and good stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a structural schematic diagram of the present invention;

[0043] Figure 2 This is a block diagram of the automatic docking system of the present invention;

[0044] Figure 3 This is a flowchart of the automatic docking process of the present invention;

[0045] Figure 4 This is a flowchart of the visual algorithm of the present invention;

[0046] Figure 5 It is a flow chart of the feature extraction algorithm of the present invention;

[0047] Figure 6 This is a structural schematic diagram of the automatic locking mechanism of the present invention.

[0048] In the figure: 1. Parallel mechanism fixed platform; 2. Parallel mechanism dynamic platform; 3. Force sensor A; 4. Force sensor B; 5. Active docking body; 6. Column A; 7. Column B; 8. Binocular camera A; 9. Binocular camera B; 10. Laser displacement sensor A; 11. Laser displacement sensor B; 12. Servo cylinder; 13. Automatic locking mechanism; 1301. Crossbar; 1302. Passive docking body; 1303. Through hole; 1304. Locking column; 1305. U-shaped connecting block; 1306. Slider; 1307. Electric slide rail. 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] See also Figure 1-6 The present invention provides a technical solution for a high-precision docking method of laser vision fusion for complex lighting environments:

[0051] according to Figure 1 、 Figure 2 and Figure 6 As shown, it includes an automatic docking system, which includes an active docking mechanism and a passive docking mechanism, and the active docking mechanism docks with the passive docking mechanism;

[0052] The active docking mechanism consists of a posture detection system, a control system, and a motion actuator. The posture detection system consists of binocular cameras A8 and B9, laser displacement sensors A10 and B11, servo cylinders 12, force sensors A3 and B4, and a vision module.

[0053] The motion actuator is composed of a parallel mechanism fixed platform 1 and a parallel mechanism dynamic platform 2;

[0054] The passive docking mechanism consists of an automatic locking mechanism 13;

[0055] The parallel mechanism fixed platform 1 is fixedly installed on the mobile platform installation surface. Force sensors A3 and B4 are installed on both sides of the parallel mechanism dynamic platform 2. Force sensors A3 and B4 support the active docking body 5. The active docking body 5 is aligned with the automatic locking mechanism 13. Columns A6 and B7 are fixedly installed on both sides of the parallel mechanism dynamic platform 2. A binocular camera A8 is installed on the top of the column A6. A binocular camera B9, a laser displacement sensor A10, and a laser displacement sensor B11 are installed on the top of the column B7. The laser displacement sensor A10 is fixed to the column B7, and one side of the laser displacement sensor B11 is connected to the output end of the servo electric cylinder 12.

[0056] The docking mechanism is divided into active docking mechanism and passive docking mechanism. The passive docking mechanism is stationary in relative space, and the active docking mechanism completes the docking action after being transported to the docking implementation area by the motion actuator. The active docking mechanism is divided into three parts: posture detection system, motion actuator, and control system. Precise docking has high requirements for the accuracy of posture detection and the stability of motion control. Multi-sensor fusion technology such as binocular camera, laser displacement sensor, force sensor, etc. is used to accurately complete the spatial 6-degree-of-freedom posture detection of the docking mechanism. A 6-degree-of-freedom parallel mechanism is used as the motion actuator. By calculating the Jacobian matrix of posture information to motion control, the posture information of the docking mechanism is mapped to the motion control quantity of the motion actuator, driving the precise docking of the active docking mechanism and the passive docking mechanism.

[0057] according to Figure 3 As shown: The high-precision docking method includes the following steps:

[0058] S1: Send the active docking mechanism to the docking implementation area and start the control system;

[0059] S2: The binocular camera B9 and the laser displacement sensor A10 start working. The binocular camera B9 acquires images, performs orientation detection and feature extraction, and the laser displacement sensor A10 acquires depth information.

[0060] S3: The visual module of the posture detection system accurately extracts the features of the passive docking mechanism within the field of view of the binocular camera B9 and integrates them with the distance information obtained by the laser displacement sensor A10. It calculates the relative posture information of the passive docking mechanism and the active docking mechanism, and converts it into the motion control quantity of the parallel mechanism dynamic platform 2 through the Jacobian matrix. The parallel mechanism dynamic platform 2 drives the active docking mechanism to approach the passive docking mechanism.

[0061] S4: When the active docking mechanism and the passive docking mechanism have reached the aligned state, the servo electric cylinder 12 pushes the laser displacement sensor B11 above the column B7 to detect and calculate the position deviation between the two. After the servo electric cylinder 12 retracts the laser displacement sensor B11, the control system drives the parallel mechanism dynamic platform 2 to drive the active docking mechanism to move and complete the docking. At the same time, the binocular camera A8 starts working. The information features of the binocular camera A8, the laser displacement sensor A10, the force sensor A3, and the force sensor B4 are extracted to determine whether there is an angular deviation between the active docking mechanism and the passive docking mechanism during the docking process, and the motion control amount of the parallel mechanism dynamic platform 2 is corrected in time.

[0062] S5: The control system sends a remote control instruction to the automatic locking mechanism 13 on the passive docking mechanism to complete the locking. After receiving the locking completion signal, the control system controls the parallel mechanism moving platform 2 to be safely retracted and removed.

[0063] according to Figure 4 As shown: The visual module encapsulates the visual algorithm and feature extraction algorithm. The visual algorithm is:

[0064] Step 1: Let the baseline length between the left and right fields of view of the binocular camera be b, the focal length be f, and the parallax of a feature point in the image between the left and right fields of view of the binocular camera be δ p , the depth value d of the feature point in the camera coordinate system can be easily obtained by the triangulation principle c for:

[0065]

[0066] Step 2: Take out three feature points obtained from the image, obtain their disparity respectively, calculate the depth information of the three points, convert the pixel coordinates to the camera coordinate system through the camera intrinsic parameters, and obtain the coordinates of the three points in the camera coordinate system, which are recorded as The extracted feature points are located on the plane of the object surface. The plane equation of the observed object surface can be obtained from the three points, which is recorded as:

[0067] aX+bY+cZ+d=0

[0068] in:

[0069]

[0070] c=(X2-X1)(Y3-Y1)-(X3-X1)(Y2-Y1)

[0071]

[0072] Step 3: According to the plane equation of the observed object surface in step 2, the installation position deviation between the laser displacement sensor and the midpoint of the binocular camera baseline in the camera coordinate system is δ=(δX cl ,δY cl ,δZ cl ), the coordinates of the surface point of the object measured by the laser displacement sensor in the camera coordinate system are p=(X p ,Y p ,Z p ), where X p =δX cl , Y p =δY cl , so point p=(δX cl ,δY cl ,Z p );

[0073] Step 4: Substitute point p into the plane equation:

[0074] aδX cl +bδY cl +cZ p +d=0

[0075] Get the depth of point p in the camera coordinate system:

[0076]

[0077] Step 5: Note that the depth measured by the laser displacement sensor at point p is d l , the same distance estimated by the binocular camera depth is d c ,in:

[0078] d c =Z p -δZ cl

[0079] Let the depth confidence of the feature points estimated by the binocular camera be c, then:

[0080]

[0081] If the depth confidence c of the feature point is higher than the set threshold, the coordinates of the feature point are used to calculate the relative position and posture of the docking mechanism. If the depth confidence c of the feature point is lower than the set threshold, the data obtained this time is discarded. When the binocular camera on one side cannot extract enough information after multiple attempts, the camera on the other side is started to perform feature extraction. For outdoor working environments, due to the directionality of light, when the lighting environment in the field of view of the camera on one side is relatively poor, the field of view of the camera on the other side can often obtain a better lighting environment. The light spots caused by the indoor lighting environment are often small in size, with obvious features and easy to eliminate. Therefore, this visual algorithm solution can adapt well to working environments with complex lighting.

[0082] according to Figure 5 As shown: The feature extraction algorithm includes the following steps:

[0083] Step 1: Grayscale and distortion correct the acquired image as the original input;

[0084] Step 2: Based on the characteristics of the image in the field of view, the average pixel gradient of the full frame is obtained in the direction with less interference;

[0085] Step 3: Based on the pre-set window size of the maximum tilt angle of the docking mechanism, with the range of the rough boundary as the center, find the maximum gradient point of each row of pixels, perform straight line fitting on these points, and obtain the approximate boundary line equation;

[0086] Step 4: Divide the window into a new window according to the new line equation, and set the horizontal boundary according to the slope of the boundary to find the required sliding window height. Slide the sliding window along the extracted vertical boundary in this area, and find the upper and lower boundaries of the end row by row.

[0087] Step 5: Divide the window within the range, eliminate the interference of background objects, and find the maximum pixel gradient point by column. Fit the obtained points with a straight line to obtain the upper and lower boundaries;

[0088] Step 6: Calculate the intersection of the four obtained boundaries to obtain the four vertices at the end of the docking mechanism for the next pose calculation.

[0089] This feature extraction algorithm replaces the original one developed by using the general feature extraction algorithm in open source visual libraries such as OpenCV. The feature points obtained are very unstable and it is difficult to distinguish the target feature from the boundaries and vertices of the end of the docking mechanism. This feature extraction algorithm has fast speed, high accuracy, good stability, and strong anti-interference ability, which can meet the technical requirements of precise docking.

[0090] according to Figure 6 As shown: the automatic locking mechanism 13 includes a cross bar 1301, a passive docking body 1302, a lock column 1304, a U-shaped connecting block 1305, a slider 1306 and an electric slide rail 1307. The passive docking body 1302 is fixedly installed on one side of the bottom end of the cross bar 1301, and a through hole 1303 is opened on one side of the passive docking body 1302. The electric slide rail 1307 is fixedly installed on the other side of the bottom end of the cross bar 1301. The surface of the electric slide rail 1307 is slidably connected with the slider 1306. The bottom end of the slider 1306 is fixedly installed with a U-shaped connecting block 1305. The bottom of the U-shaped connecting block 1305 is installed with a lock column 1304, and one end of the lock column 1304 is plugged into the through hole 1303.

[0091] The operation of the electric slide rail 1307 is controlled by a control system, the electric slide rail 1307 drives the slider 1306 to move, the movement of the slider 1306 drives the U-shaped connecting block 1305 to move, and the movement of the U-shaped connecting block 1305 drives the lock column 1304 to move, so that the lock column 1304 is inserted into the through hole 1303 for self-locking.

[0092] Working principle: The present invention provides a high-precision docking method for laser vision fusion in complex lighting environments. First, the active docking mechanism is transported to the docking implementation area by the motion actuator, the control system is started, and the binocular camera B9 above the column B7 and the fixed laser displacement sensor A10 start working. The visual module of the posture detection system accurately extracts the characteristics of the passive docking mechanism within the field of view of the binocular camera B9, and fuses it with the distance information obtained by the laser displacement sensor A10 to calculate the relative posture information of the passive docking mechanism and the active docking mechanism. The information is converted into the motion control quantity of the motion actuator through the Jacobian matrix, and the motion actuator is driven to drive the active docking mechanism to approach the passive docking mechanism. When the approach action is completed, the active docking mechanism and the passive docking mechanism have reached an alignment state. At this time, the servo electric cylinder 12 pushes the active laser displacement sensor B above the column B7. 11, detect and calculate the position deviation between the two, after the servo electric cylinder 12 retracts the active laser displacement sensor B11, the control system drives the motion actuator to drive the active docking mechanism to move to complete the docking process. During the docking process, the binocular camera A8 on the column A6 starts working, and judges whether there is an angle deviation between the active and passive docking mechanisms during the docking process through feature extraction, and promptly corrects the motion control amount of the parallel mechanism. In addition, the control system monitors the data of the laser displacement sensor B11 to ensure that the docking is accurate and safe, and there will be no damage to the docking mechanism due to data errors. After the docking is completed, the control system sends a remote control instruction to the automatic locking mechanism 13 on the passive docking mechanism to complete the locking. After receiving the locking completion signal, the motion actuator is controlled to safely retract the parallel mechanism moving platform 2 and move the parallel mechanism moving platform 2 away.

[0093] In the description of the present invention, it should be understood that the indicated orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0094] In the present invention, unless otherwise clearly specified and limited, for example, it can be a fixed connection, a detachable connection, or an integrated connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, it can be internal communication between two elements or an interaction relationship between two elements. Unless otherwise clearly specified and limited, ordinary technicians in this field can understand the specific meanings of the above terms in the present invention according to specific circumstances.

[0095] While 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 these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A high-precision docking method for laser vision fusion in complex lighting environments, characterized in that: It includes an automatic docking system, the automatic docking system includes an active docking mechanism and a passive docking mechanism, the active docking mechanism docks with the passive docking mechanism; The active docking mechanism comprises a posture detection system, a control system and a motion execution mechanism, wherein the posture detection system comprises a binocular camera A (8), a binocular camera B (9), a laser displacement sensor A (10), a laser displacement sensor B (11), a servo electric cylinder (12), a force sensor A (3), a force sensor B (4) and a vision module, wherein the motion execution mechanism comprises a parallel mechanism fixed platform (1) and a parallel mechanism moving platform (2), and the passive docking mechanism comprises an automatic locking mechanism (13), wherein the parallel mechanism fixed platform (1) is fixedly mounted on the moving platform mounting surface, and the force sensor A (3) and the force sensor B (4) are respectively mounted on both sides of the parallel mechanism moving platform (2). Device B (4), the force sensor A (3) and the force sensor B (4) support the active docking body (5), the active docking body (5) is aligned with the automatic locking mechanism (13), and the two sides of the parallel mechanism dynamic platform (2) are respectively fixedly mounted with a column A (6) and a column B (7), the top of the column A (6) is mounted with a binocular camera A (8), the top of the column B (7) is mounted with a binocular camera B (9), a laser displacement sensor A (10), and a laser displacement sensor B (11), the laser displacement sensor A (10) is fixed to the column B (7), and one side of the laser displacement sensor B (11) is connected to the output end of the servo electric cylinder (12); The high-precision docking method comprises the following steps: S1: Send the active docking mechanism to the docking implementation area and start the control system; S2: The binocular camera B (9) and the laser displacement sensor A (10) start working. The binocular camera B (9) acquires images, performs orientation detection and feature extraction, and the laser displacement sensor A (10) acquires depth information. S3: The visual module of the posture detection system accurately extracts the features of the passive docking mechanism within the field of view of the binocular camera B (9) and fuses them with the distance information obtained by the laser displacement sensor A (10), calculates the relative posture information of the passive docking mechanism and the active docking mechanism, and converts it into the motion control quantity of the parallel mechanism dynamic platform (2) through the Jacobian matrix, driving the parallel mechanism dynamic platform (2) to drive the active docking mechanism to approach the passive docking mechanism; S4: When the active docking mechanism and the passive docking mechanism have reached an aligned state, the servo electric cylinder (12) pushes the laser displacement sensor B (11) above the column B (7) to detect and calculate the position deviation between the two. After the servo electric cylinder (12) retracts the laser displacement sensor B (11), the control system drives the parallel mechanism moving platform (2) to drive the active docking mechanism to move and complete the docking. At the same time, the binocular camera A (8) starts working. The information features of the binocular camera A (8), the laser displacement sensor A (10), the force sensor A (3), and the force sensor B (4) are extracted to determine whether an angle deviation occurs between the active docking mechanism and the passive docking mechanism during the docking process, and the motion control amount of the parallel mechanism moving platform (2) is corrected in time. S5: The control system sends a remote control command to the automatic locking mechanism (13) on the passive docking mechanism to complete the locking. After receiving the signal of the locking completion, the control system controls the parallel mechanism moving platform (2) to be safely retracted and removed.

2. The high-precision docking method for laser vision fusion in complex lighting environments according to claim 1, characterized in that: The visual module encapsulates the visual algorithm and feature extraction algorithm, and the visual algorithm is: Step 1: Let the baseline length between the left and right fields of view of the binocular camera be b, the focal length be f, and the parallax of a feature point in the image between the left and right fields of view of the binocular camera be δ p , the depth value d of the feature point in the camera coordinate system can be easily obtained by the triangulation principle c for: Step 2: Take out three feature points obtained from the image, obtain their disparity respectively, calculate the depth information of the three points, convert the pixel coordinates to the camera coordinate system through the camera intrinsic parameters, and obtain the coordinates of the three points in the camera coordinate system, which are recorded as The extracted feature points are located on the plane of the object surface. The plane equation of the observed object surface can be obtained from the three points, which is recorded as: aX+bY+cZ+d=0 in: c=(X2-X1)(Y3-Y1)-(X3-X1)(Y2-Y1) Step 3: According to the plane equation of the observed object surface in step 2, the installation position deviation between the laser displacement sensor and the midpoint of the binocular camera baseline in the camera coordinate system is δ=(δX cl ,δY cl ,δZ cl ), the coordinates of the surface point of the object measured by the laser displacement sensor in the camera coordinate system are p=(X p ,Y p ,Z p ), where X p =δX cl , Y p =δY cl , so point p=(δX cl ,δY cl ,Z p ); Step 4: Substitute point p into the plane equation: aδX cl +bδY cl +cZ p +d=0 Get the depth of point p in the camera coordinate system: Step 5: Note that the depth measured by the laser displacement sensor at point p is d l , the same distance estimated by the binocular camera depth is d c ,in: d c =Z p -δZ cl Let the depth confidence of the feature points estimated by the binocular camera be c, then:

3. The high-precision docking method for laser vision fusion in complex lighting environments according to claim 2, characterized in that: If the depth confidence c of the feature point is higher than the set threshold, the coordinates of the feature point are used to calculate the relative position and posture of the docking mechanism. If the depth confidence c of the feature point is lower than the set threshold, the data obtained this time is discarded.

4. The high-precision docking method for laser vision fusion in complex lighting environments according to claim 2, characterized in that: The feature extraction algorithm includes the following steps: Step 1: Grayscale and distortion correct the acquired image as the original input; Step 2: Based on the characteristics of the image in the field of view, the average pixel gradient of the full frame is obtained in the direction with less interference; Step 3: Based on the pre-set window size of the maximum tilt angle of the docking mechanism, with the range of the rough boundary as the center, find the maximum gradient point of each row of pixels, perform straight line fitting on these points, and obtain the approximate boundary line equation; Step 4: Divide the window into a new window according to the new line equation, and set the horizontal boundary according to the slope of the boundary to find the required sliding window height. Slide the sliding window along the extracted vertical boundary in this area, and find the upper and lower boundaries of the end row by row. Step 5: Divide the window within the range, eliminate the interference of background objects, and find the maximum pixel gradient point by column. Fit the obtained points with a straight line to obtain the upper and lower boundaries; Step 6: Calculate the intersection of the four obtained boundaries to obtain the four vertices at the end of the docking mechanism for the next pose calculation.

5. The high-precision docking method for laser vision fusion in complex lighting environments according to claim 1, characterized in that: The automatic locking mechanism (13) comprises a cross bar (1301), a passive docking body (1302), a locking column (1304), a U-shaped connecting block (1305), a slider (1306) and an electric slide rail (1307). The passive docking body (1302) is fixedly mounted on one side of the bottom end of the cross bar (1301). A through hole (1303) is provided on one side of the passive docking body (1302). The electric slide rail (1307) is fixedly mounted on the other side of the bottom end of the cross bar (1301). The surface of the electric slide rail (1307) is slidably connected to the slider (1306). The bottom end of the slider (1306) is fixedly mounted with a U-shaped connecting block (1305). The bottom of the U-shaped connecting block (1305) is mounted with a locking column (1304). One end of the locking column (1304) is plugged into the through hole (1303).

Images