A robot 3D vision positioning system and method based on area array TOF sensing guidance

Abstract

The application belongs to the technical field of industrial robot automation processing, and particularly relates to a robot 3D vision positioning system and method based on area array TOF sensing guidance, which comprises a 3D vision camera and an area array TOF sensor installed at the end of a mechanical arm, and a control system. The area array TOF sensor collects distance data of multiple independent sensing areas in its field of view range at one time to form a local distance distribution matrix. The control system calculates the spatial posture of the local surface of a workpiece according to the matrix, including a normal vector and a curvature, and cooperatively adjusts the position and posture of the robot, so that the end effector meets the distance and angle requirements at the same time when approaching the workpiece; meanwhile, intelligent guidance and collision warning of a region of interest are supported. The application solves the problem that the existing coarse positioning stage lacks local spatial form perception, which leads to the difficulty of 3D camera scanning in the best posture, significantly improves the first scanning success rate and operation efficiency, and can be independently applied to a welding robot or a polishing robot.

Description

Technical Field

[0001] This invention belongs to the field of industrial robot automated processing technology, specifically a robot 3D vision positioning system and method based on area array TOF perception guidance. Background Technology

[0002] In robotic automated welding and grinding operations, utilizing 3D vision systems for workpiece identification and positioning is a key technology for achieving "teach-free" flexible machining. However, existing technologies have the following shortcomings in positioning and guiding complex workpieces:

[0003] 1. Limitations of single-point ranging: Existing assisted positioning solutions mostly use single-point lasers or single-point TOF sensors, which can only acquire "distance" information and cannot perceive the local pose (normal vector, curvature), edge position, and multi-target distribution of the workpiece surface. When dealing with complex curved workpieces such as bent plates and pipes, even if the distance meets the requirements, the angle between the optical axis of the 3D camera and the workpiece surface may be too large, resulting in imaging distortion and point cloud distortion.

[0004] 2. The "blind men and the elephant" approach process: Traditional coarse positioning can only tell "how far away from the workpiece," but cannot tell "in which direction the workpiece is tilted" or "which area is the target area that needs to be scanned in detail." When the robot approaches a complex workpiece, it lacks knowledge of the local geometry and is prone to approaching the wrong parts or colliding with protrusions.

[0005] 3. Low success rate of first scan: For complex workpieces with multiple features and deep and narrow bevels, the 3D camera needs to adjust its pose multiple times to obtain effective scan data, which seriously affects the operation cycle.

[0006] 4. Special challenges in grinding scenarios: The surface of the workpiece being ground has raised features such as weld seam excess and weld slag, which traditional solutions cannot detect in advance, causing the 3D camera to scan invalid areas and increasing the risk of collision when approaching.

[0007] Therefore, there is an urgent need for an auxiliary positioning scheme capable of sensing the local spatial morphology of a workpiece to accurately guide the 3D camera to complete the scan at the optimal position and orientation. This invention can be applied independently to welding robots or grinding robots. Summary of the Invention

[0008] The purpose of this invention is to provide a robot 3D vision positioning system and method based on area array TOF perception guidance, in order to solve the technical problem that the lack of workpiece local spatial shape perception capability in the coarse positioning stage makes it difficult for 3D vision cameras to complete high-precision scanning of complex workpieces with optimal pose.

[0009] The technical solution adopted by this invention to achieve the above objectives is: a robot 3D vision positioning system based on area array TOF perception guidance, comprising:

[0010] A 3D vision camera, mounted at the end of the robotic arm of the robot, is used to scan and obtain the point cloud of the workpiece;

[0011] The area array TOF sensor is installed at the end of the robotic arm of the robot body and is fixed relative to the 3D vision camera. It is used to collect distance data from multiple independent sensing areas within its field of view to the workpiece surface at one time, forming a local distance distribution matrix.

[0012] The communication module is used to communicate with the robot body and the 3D vision camera;

[0013] The control system is used to calculate the spatial orientation of the local surface of the workpiece based on the local distance distribution matrix, and coordinate the adjustment of the robot's position and orientation so that the end effector meets both distance and angle requirements when approaching the workpiece, and triggers the 3D vision camera to perform scanning when the preset threshold conditions are met.

[0014] The control system includes:

[0015] The spatial attitude calculation module is used to calculate the spatial attitude of the local surface of the workpiece relative to the end effector based on the local distance distribution matrix. The spatial attitude includes at least local normal vectors and local curvature information.

[0016] The pose coordination control module is used to coordinately adjust the position and posture of the robot according to the calculated spatial posture; it is also used to drive the robot to adjust its pose until the region of interest is guided to the optimal working field of view center of the 3D vision camera.

[0017] The scanning trigger module is used to trigger the 3D vision camera to perform scanning when both the position and orientation meet preset threshold conditions.

[0018] The scanning area planning module is used to plan the scanning area of ​​the 3D vision camera based on the coarse point cloud data generated by the local distance distribution matrix.

[0019] The Region of Interest (ROI) configuration module is used to specify one or more of multiple independent sensing regions as ROIs, and can set the target feature type within the ROI.

[0020] The pose coordination control module calculates the offset of the centroid of the region of interest relative to the physical center of the area array TOF sensor, and drives the robot to move to compensate for the offset, so that the centroid of the region of interest is aligned with the center of the sensor.

[0021] The target feature type includes raised features or recessed features;

[0022] When the target feature type is a protrusion feature, the pose coordination control module uses the highest point of the protrusion as the guiding target, adjusts the robot position so that the highest point of the protrusion is located in the center of the field of view of the 3D vision camera, and adjusts the posture so that the angle between the average normal vector of the region of interest and the optical axis of the 3D vision camera is less than a preset angle threshold.

[0023] The area array TOF sensor includes a SPAD array composed of multiple photosensitive pixels, and its sensing area resolution is at least 4×4.

[0024] The control system further includes a collision warning module, which is used to identify protrusions or depressions on the surface of the workpiece based on the local distance distribution matrix, and issue a collision warning signal or trigger obstacle avoidance path replanning.

[0025] The collision warning module monitors the minimum distance value in the local distance distribution matrix in real time. If the minimum distance value is less than the preset safety threshold, it will pause the approach and issue a warning. At the same time, it will detect whether there is a situation where the distance difference between adjacent sensing areas exceeds the preset abrupt change threshold. If there is, it will be identified as a protrusion feature, the protrusion height will be calculated, and obstacle avoidance will be triggered.

[0026] The preset safety threshold is 30mm, and the preset mutation threshold is 20mm.

[0027] A localization method for a robot 3D vision localization system based on area array TOF perception guidance includes the following steps:

[0028] Step S1: The control system presets the optimal working distance range and optimal incident angle range for the 3D vision camera;

[0029] Step S2: The robot moves to the initial position above the workpiece and activates the area array TOF sensor;

[0030] Step S3: The control system acquires the local distance distribution matrix through the area array TOF sensor;

[0031] Step S4: The control system calculates the local normal vector of the local surface of the workpiece based on the local distance distribution matrix;

[0032] Step S5: The control system plans the scanning area of ​​the 3D vision camera based on the local distance distribution matrix;

[0033] Step S6: The control system determines whether the current working distance meets the optimal working distance range and whether the angle between the optical axis of the 3D vision camera and the local normal vector meets the optimal incident angle range; if not, the control system coordinates to adjust the position and posture of the robot to gradually approach the optimal value; if it meets the requirements, proceed to step S7.

[0034] Step S7: Control the system to trigger the 3D vision camera to scan and acquire high-precision point cloud;

[0035] Step S8: The control system identifies target features based on the high-precision point cloud, plans the processing path, and controls the robot to perform the operation.

[0036] In step S1, when used for welding operations: the optimal working distance range is 400±50mm, and the optimal incident angle range is 0±15°; wherein, 400±50mm and 0±15° correspond to the working parameters of the 3D vision camera imaging distortion rate being less than 5%.

[0037] Region of Interest (ROI) guidance steps prior to step S3:

[0038] The operator selects a continuous region as the region of interest on the virtual grid of the area array TOF sensor by row and column, and sets the target feature type as a raised feature or a recessed feature.

[0039] The control system calculates the centroid coordinates of the region of interest. Compared to the physical center of an array-type TOF sensor The offset is:

[0040]

[0041] Drive the robot to move along the X and Y directions and Compensation is performed to align the centroid of the region of interest with the center of the sensor, thereby guiding it to the center of the field of view of the 3D vision camera;

[0042] When the target feature type is a convex feature, the point with the minimum distance in the local distance distribution matrix is ​​taken as the highest point of the convex feature, and the robot position is adjusted so that the point is located in the center of the field of view of the 3D vision camera.

[0043] In step S5, the scanning area of ​​the 3D vision camera is planned based on the local distance distribution matrix, specifically as follows:

[0044] Step S5-1: The control system will use the local distance distribution matrix Convert to 3D coarse point cloud ,in , For the first Line number The physical coordinates corresponding to the column sensing area;

[0045] Step S5-2: Process the coarse point cloud Statistical filtering is performed to remove outliers, resulting in an effective coarse point cloud. ;

[0046] Step S5-3: Calculation The convex hull or minimum bounding rectangle on the horizontal plane is denoted as the boundary. ;

[0047] Step S5-4: Set the boundary Extend outward by the preset margin distance Generate the boundary of the scanned region ,in The diameter is 5mm to 20mm;

[0048] Step S5-5: Define the boundaries of the scanned area As the scanning field of view of the 3D vision camera, it drives the robot to move so that the range is within the field of view of the 3D vision camera.

[0049] In step S6, when coordinating the adjustment of position and attitude:

[0050] The control system is based on the local distance distribution matrix. ,in, Let be the distance value of the sensing area in the i-th row and j-th column, i=1-m, j=1-n, where m and n are the row and column numbers, respectively. Calculate the average distance μ, i.e.:

[0051]

[0052] The local normal vector is calculated by fitting the local plane using the least squares method. for:

[0053]

[0054] Let v be the direction vector of the optical axis of the 3D vision camera, then the included angle θ is:

[0055]

[0056] like Then, control the robot to rotate around the X and Y axes of the tool coordinate system, with the rotation angle increment being... ,in, This is a proportionality coefficient, with a value ranging from 0.1 to 0.5.

[0057] If μ exceeds the distance range, the robot moves along the optical axis, with a step size of [missing value]. ,in, This is the distance adjustment factor, with a value ranging from 0.2 to 0.8.

[0058] The present invention has the following beneficial effects and advantages:

[0059] 1. This invention is the first to calculate the local normal vector and curvature of the workpiece in the coarse positioning stage, ensuring that the 3D camera scans at the optimal incident angle and avoiding imaging distortion.

[0060] 2. This invention ensures that the weld seam or excess height area is accurately guided to the center of the camera's field of view, significantly improving the success rate of the first scan. Experimental data shows that the success rate has increased from less than 70% to over 95%.

[0061] 3. This invention utilizes spatial resolution capabilities to identify protruding features in advance, thereby achieving active obstacle avoidance.

[0062] 4. This invention utilizes TOF coarse point cloud to plan the scanning area, reducing the amount of data by more than 90% and shortening the time by more than 80%. Attached Figure Description

[0063] Figure 1 A schematic diagram of the system structure of this invention;

[0064] Figure 2 A schematic diagram of the sensing area of ​​an 8x8 area array TOF sensor according to an embodiment of the present invention;

[0065] Figure 3 Flowchart of the 3D visual positioning method of the present invention;

[0066] Figure 4 A schematic diagram of bending plate posture adjustment in a welding scenario according to an embodiment of the present invention;

[0067] Figure 5 This invention provides a schematic diagram of the ROI (Residual Area of ​​Interest) guidance for weld seam reinforcement in a grinding scenario. Detailed Implementation

[0068] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. Four specific embodiments of the present invention are respectively applied to different scenarios of welding robots and grinding robots. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0069] like Figure 1 As shown, the present invention discloses a positioning method for a robot 3D vision positioning system based on area array TOF perception guidance, comprising the following steps:

[0070] Step S1: The control system presets the optimal working distance range and optimal incident angle range for the 3D vision camera.

[0071] Step S2: The robot moves to the initial position above the workpiece and activates the area array TOF sensor.

[0072] Step S3: The control system acquires the local distance distribution matrix through the area array TOF sensor.

[0073] Prior to step S3, a region of interest guidance step may also be included:

[0074] The operator selects a continuous region as the region of interest on the virtual grid of the area array TOF sensor and sets the target feature type as a raised feature or a recessed feature.

[0075] The control system calculates the centroid coordinates of the region of interest. Compared to the physical center of an array-type TOF sensor offset , And drive the robot to move along the X and Y directions. and Compensation is performed to align the centroid of the region of interest with the center of the sensor;

[0076] When the target feature type is a convex feature, the point with the minimum distance in the local distance distribution matrix is ​​taken as the highest point of the convex feature, and the robot position is adjusted so that the point is located in the center of the field of view of the 3D vision camera.

[0077] Step S4: The control system calculates the local normal vector of the local surface of the workpiece based on the local distance distribution matrix.

[0078] Step S5: The control system plans the scanning area of ​​the 3D vision camera based on the local distance distribution matrix, specifically as follows:

[0079] Step S5-1: Calculate the local distance distribution matrix Convert to 3D coarse point cloud ,in, , For the first Line number The physical coordinates corresponding to the column sensing area;

[0080] Step S5-2: Process the coarse point cloud Statistical filtering is performed to remove outliers, resulting in an effective coarse point cloud. ;

[0081] Step S5-3: Calculation The convex hull or minimum bounding rectangle on the horizontal plane is denoted as the boundary. ;

[0082] Step S5-4: Set the boundary Extend outward by the preset margin distance Generate the boundary of the scanned region ,in, The diameter is 5mm to 20mm;

[0083] Step S5-5: Define the boundaries of the scanned area As the scanning field of view of the 3D vision camera, it drives the robot to move so that the range is within the field of view of the 3D vision camera.

[0084] Step S6: The control system determines whether the current working distance meets the optimal working distance range and whether the angle between the optical axis of the 3D vision camera and the local normal vector meets the optimal incident angle range; if not, the control system coordinates to adjust the position and posture of the robot to gradually approach the optimal value; if it meets the requirements, proceed to step S7.

[0085] Specifically, when coordinating the adjustment of position and attitude:

[0086] Based on the local distance distribution matrix Calculate the mean distance:

[0087] ;

[0088] The local normal vector is calculated by fitting the local plane using the least squares method. ;

[0089] Let the optical axis direction vector of the 3D vision camera be... Then the included angle ;

[0090] like Then, control the robot to rotate around the X and Y axes of the tool coordinate system, with the rotation angle increment being... ,in, The value range is 0.1 to 0.5;

[0091] like If the distance exceeds the specified range, the robot will move along the optical axis, with a step size of [missing information]. ,in, The value range is 0.2 to 0.8.

[0092] Step S7: The control system triggers the 3D vision camera to scan and acquire high-precision point clouds.

[0093] Step S8: The control system identifies target features based on the high-precision point cloud, plans the processing path, and controls the robot to perform the operation.

[0094] In addition, the control system of the present invention also includes a collision warning module. This module monitors the minimum distance value in the local distance distribution matrix in real time. If the minimum distance value is less than a preset safety threshold of 30mm, the approach is paused and a warning is issued. At the same time, it detects whether there is a situation where the distance difference between adjacent sensing areas exceeds a preset abrupt change threshold of 20mm. If so, it is identified as a protrusion feature, the protrusion height is calculated, and obstacle avoidance path replanning is triggered.

[0095] The method steps described above are explained below with reference to specific embodiments. The four specific embodiments of the present invention are respectively applied to different scenarios of welding robots and grinding robots. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0096] Example 1: Welding Application—Adaptive Orientation Guidance for Complex Curved Workpieces

[0097] like Figure 1 As shown, this embodiment takes a welding robot as an example to demonstrate how a TOF area array sensor can acquire the local normal vector of the workpiece and guide the robot to complete adaptive posture adjustment for complex workpieces such as bent plates in shipbuilding.

[0098] System configuration: a six-axis welding robot; a 3D camera with an optimal working distance of 400±50mm and an incident angle of 0±15° (corresponding to an imaging distortion rate of less than 5%); an 8×8 resolution area array TOF sensor (such as VL53L8CX), mounted at the end of the robotic arm and fixed relative to the 3D vision camera. Figure 1 As shown.

[0099] Workpiece characteristics: Curved plate with a radius of curvature of approximately 500 mm.

[0100] like Figure 3 , Figure 4 As shown, the working process is as follows: The robot moves to a position approximately 600mm above the workpiece, and the control system activates the area array TOF sensor to collect distance data from 64 sensing areas at once, forming a local distance distribution matrix M. The system identifies a central region distance of approximately 580mm and an edge distance of approximately 520mm, exhibiting a distribution with a convex center and concave edges. The spatial pose calculation module performs planar fitting on M, solving for the local normal vector n. The calculated angle θ between the current 3D camera optical axis and n is 25°, exceeding the 15° threshold. Figure 4 As shown, the pose coordination control module issues a command, and while the robot moves downward, it rotates approximately 10° around the horizontal axis in the workpiece tilt direction, monitoring the average distance μ and the included angle θ in real time. When μ drops to 410mm and θ drops to 12°, the pose is locked, and the scanning trigger module triggers the 3D camera to scan, acquiring a complete and distortion-free point cloud of the curved plate in one go.

[0101] Technical advantages: Traditional single-point TOF solutions can only adjust distance and cannot sense attitude. The incident angle remains at 25°, leading to point cloud distortion that requires secondary adjustments. This embodiment achieves highly efficient positioning with "one approach, one successful scan" through attitude sensing.

[0102] Example 2: Welding Application—Intelligent ROI Guidance for Complex Workpieces with Multiple Welds

[0103] like Figures 2-3As shown, this embodiment uses a welding robot as an example to demonstrate the intelligent guidance function of the region of interest (ROI) for large structural components with multiple weld seam features.

[0104] Workpiece characteristics: Box-shaped structural component, the target weld is located in a groove about 30mm deep, with raised structures around it.

[0105] like Figures 2-3 As shown, the working process of this embodiment is as follows: The operator, on the 8×8 virtual grid of the control system, such as... Figure 2 As shown, the central area (rows 3-6 and columns 3-6) is selected as the Region of Interest (ROI), corresponding to the weld seam within the groove, and the target feature type is set. The robot moves above the workpiece, and Time-of-Flight (TOF) acquires distance values. The system detects that the ROI's distance value is greater than the surrounding area (groove features), and the ROI's centroid is approximately 15mm off-center from the sensor's physical center. The ROI guidance module calculates the offset (Δx=15mm, Δy=8mm), and the pose coordination control module drives the robot to move 12mm in the X direction and 8mm in the Y direction, aligning the ROI's centroid with the sensor center. During the approach, the average ROI distance is used as the primary control variable. When the average ROI distance drops to 390mm and the standard deviation of this distance is <5mm (the bottom of the groove is essentially parallel to the sensor plane), a 3D camera scan is triggered to clearly capture the weld seam features at the bottom of the groove, unaffected by surrounding protrusions. Figure 3 As shown, the above process corresponds to steps S3-S7 in the overall flowchart of the method.

[0106] Technical effect: Traditional global positioning is easily interfered with by protrusions. This embodiment improves the success rate of the first scan from less than 70% to more than 95% by focusing on the ROI.

[0107] Example 3: Grinding Application—Intelligent Guidance and Positioning of Weld Residual Height Area

[0108] like Figure 5 As shown, this embodiment takes a grinding robot as an example to demonstrate how to use a surface array TOF to identify protruding features and guide a 3D camera to scan accurately for grinding and positioning of the weld seam after welding.

[0109] System configuration: Six-axis grinding robot, 3D camera with optimal working distance of 300±30mm, incident angle of 0±10°, and preset collision safety distance of 30mm.

[0110] Workpiece characteristics: Butt weld of steel plate, with a reinforcement height of 3-5mm and a width of 15-20mm.

[0111] The working process of this embodiment is as follows: The operator selects the ROI (rows 2-7, columns 3-6) and sets the target feature as "protrusion (weld excess height)". The system uses the point with the minimum distance in the local distance distribution matrix as the highest point of the protrusion. The robot moves to about 500mm above the workpiece, and TOF collects distance values. The system analysis shows that the distance in the area of ​​rows 4-5 and columns 3-4 is about 4mm smaller than the surrounding area (local protrusion feature). The attitude calculation module calculates the local curvature κ>0 within the ROI area, confirming it as a positive protrusion (weld excess height). The pose collaborative control module uses the highest point of the protrusion as the guiding target, adjusts the robot position so that this point is located in the center of the camera's field of view, and adjusts the attitude so that the angle between the average normal vector of the ROI area and the optical axis is <8°. Figure 5 As shown, a 3D camera scan is triggered to acquire the complete 3D contour (height, width, and length) of the excess height in one go. The control system plans the grinding path (pressure, feed rate, and number of reciprocations) based on the excess height model and executes automatic grinding.

[0112] Technical effect: The area array TOF actively identifies convex features, guides the 3D camera to scan accurately, and reduces the positioning time by about 70%.

[0113] Example 4: Grinding Application—Collision Warning and Safe Approach (Identifying Weld Slag Protrusions)

[0114] like Figure 1 and Figure 3 As shown in the figure, this embodiment takes a grinding robot as an example to demonstrate how the collision warning module uses the spatial resolution capability of the area array TOF to identify protrusions such as welding slag and spatter, and achieve active obstacle avoidance.

[0115] Workpiece characteristics: After welding, the structural parts have irregular weld slag protrusions (8-12mm in height) on the surface.

[0116] The working process of this embodiment: The robot approaches the workpiece at a medium speed, and Time-of-Flight (TOF) continuously collects distance data from 64 regions, updating the local distance distribution matrix in real time. The collision warning module detects a distance value of 185mm in the 2nd row and 7th column, with adjacent regions having distance values ​​of 260-280mm and a depth difference of 75-95mm, far exceeding the preset abrupt change threshold (20mm). The module identifies this anomaly as a weld slag protrusion, calculating its height to be approximately 10mm. Due to the high-speed rotation of the grinding head, a high risk is identified, triggering an immediate warning. The robot pauses its movement, and the control system highlights the location of the weld slag. Figure 1 As shown. After operator confirmation, the system invokes obstacle avoidance path replanning to bypass the protrusion. Figure 3 During the approach process in steps S5-S6, the collision warning module can be activated for real-time monitoring.

[0117] Technical benefits: Traditional single-point Time-of-Flight (TOF) sensors cannot detect the spatial location of protrusions, which can easily lead to collisions. The area array TOF of this invention has spatial resolution capabilities, enabling active obstacle avoidance and ensuring the safety of equipment and workpieces.

[0118] In summary, in conjunction with the above embodiments, this invention achieves local spatial morphology recognition of workpieces in the coarse positioning stage through area array TOF perception guidance, effectively solving the problem of insufficient information in traditional single-point ranging, significantly improving the first scan success rate and operation efficiency of 3D visual positioning, and also has active collision warning capability, making it suitable for various robotic automated processing scenarios such as welding and grinding.

[0119] Those skilled in the art will understand that the above description is merely a preferred embodiment of the present invention, and the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. This is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0120] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A robot 3D vision positioning system based on area array TOF perception guidance, characterized in that, include: A 3D vision camera, mounted at the end of the robotic arm of the robot, is used to scan and obtain the point cloud of the workpiece; The area array TOF sensor is installed at the end of the robotic arm of the robot body and is fixed relative to the 3D vision camera. It is used to collect distance data from multiple independent sensing areas within its field of view to the workpiece surface at one time, forming a local distance distribution matrix. The communication module is used to communicate with the robot body and the 3D vision camera; The control system is used to calculate the spatial orientation of the local surface of the workpiece based on the local distance distribution matrix, and coordinate the adjustment of the robot's position and orientation so that the end effector meets both distance and angle requirements when approaching the workpiece, and triggers the 3D vision camera to perform scanning when the preset threshold conditions are met.

2. The robot 3D vision positioning system based on area array TOF perception guidance according to claim 1, characterized in that, The control system includes: The spatial attitude calculation module is used to calculate the spatial attitude of the local surface of the workpiece relative to the end effector based on the local distance distribution matrix. The spatial attitude includes at least local normal vectors and local curvature information. The pose coordination control module is used to coordinately adjust the position and posture of the robot according to the calculated spatial posture; it is also used to drive the robot to adjust its pose until the region of interest is guided to the optimal working field of view center of the 3D vision camera. The scanning trigger module is used to trigger the 3D vision camera to perform scanning when both the position and orientation meet preset threshold conditions. The scanning area planning module is used to plan the scanning area of ​​the 3D vision camera based on the coarse point cloud data generated by the local distance distribution matrix. The Region of Interest (ROI) configuration module is used to specify one or more of multiple independent sensing regions as ROIs, and can set the target feature type within the ROI. The pose coordination control module calculates the offset of the centroid of the region of interest relative to the physical center of the area array TOF sensor, and drives the robot to move to compensate for the offset, so that the centroid of the region of interest is aligned with the center of the sensor.

3. The robot 3D vision positioning system based on area array TOF perception guidance according to claim 2, characterized in that, The target feature type includes raised features or recessed features; When the target feature type is a protrusion feature, the pose coordination control module uses the highest point of the protrusion as the guiding target, adjusts the robot position so that the highest point of the protrusion is located in the center of the field of view of the 3D vision camera, and adjusts the posture so that the angle between the average normal vector of the region of interest and the optical axis of the 3D vision camera is less than a preset angle threshold.

4. The robot 3D vision positioning system based on area array TOF perception guidance according to claim 1, characterized in that, The area array TOF sensor includes a SPAD array composed of multiple photosensitive pixels, and its sensing area resolution is at least 4×4.

5. A robot 3D vision positioning system based on area array TOF perception guidance according to claim 1, characterized in that, The control system further includes a collision warning module, which is used to identify protrusions or depressions on the surface of the workpiece based on the local distance distribution matrix, and issue a collision warning signal or trigger obstacle avoidance path replanning. The collision warning module monitors the minimum distance value in the local distance distribution matrix in real time. If the minimum distance value is less than the preset safety threshold, it will pause the approach and issue a warning. At the same time, it will detect whether there is a situation where the distance difference between adjacent sensing areas exceeds the preset abrupt change threshold. If there is, it will be identified as a protrusion feature, the protrusion height will be calculated, and obstacle avoidance will be triggered. The preset safety threshold is 30mm, and the preset mutation threshold is 20mm.

6. The positioning method of a robot 3D vision positioning system based on area array TOF perception guidance according to claim 1, characterized in that, Includes the following steps: Step S1: The control system presets the optimal working distance range and optimal incident angle range for the 3D vision camera; Step S2: The robot moves to the initial position above the workpiece and activates the area array TOF sensor; Step S3: The control system acquires the local distance distribution matrix through the area array TOF sensor; Step S4: The control system calculates the local normal vector of the local surface of the workpiece based on the local distance distribution matrix; Step S5: The control system plans the scanning area of ​​the 3D vision camera based on the local distance distribution matrix; Step S6: The control system determines whether the current working distance meets the optimal working distance range and whether the angle between the optical axis of the 3D vision camera and the local normal vector meets the optimal incident angle range; if not, the control system coordinates to adjust the position and posture of the robot to gradually approach the optimal value; if it meets the requirements, proceed to step S7. Step S7: Control the system to trigger the 3D vision camera to scan and acquire high-precision point cloud; Step S8: The control system identifies target features based on the high-precision point cloud, plans the processing path, and controls the robot to perform the operation.

7. The positioning method of a robot 3D vision positioning system based on area array TOF perception guidance according to claim 6, characterized in that, In step S1, when used for welding operations: the optimal working distance range is 400±50mm, and the optimal incident angle range is 0±15°; wherein, 400±50mm and 0±15° correspond to working parameters where the imaging distortion rate of the 3D vision camera is less than 5%.

8. The positioning method of a robot 3D vision positioning system based on area array TOF perception guidance according to claim 6, characterized in that, Region of Interest (ROI) guidance steps prior to step S3: The operator selects a continuous region as the region of interest on the virtual grid of the area array TOF sensor by row and column, and sets the target feature type as a raised feature or a recessed feature. The control system calculates the centroid coordinates of the region of interest. Compared to the physical center of an array-type TOF sensor The offset is: ； Drive the robot to move along the X and Y directions and Compensation is performed to align the centroid of the region of interest with the center of the sensor, thereby guiding it to the center of the field of view of the 3D vision camera; When the target feature type is a convex feature, the point with the minimum distance in the local distance distribution matrix is ​​taken as the highest point of the convex feature, and the robot position is adjusted so that the point is located in the center of the field of view of the 3D vision camera.

9. The positioning method of a robot 3D vision positioning system based on area array TOF perception guidance according to claim 6, characterized in that, In step S5, the scanning area of ​​the 3D vision camera is planned based on the local distance distribution matrix, specifically as follows: Step S5-1: The control system will use the local distance distribution matrix Convert to 3D coarse point cloud ,in , For the first Line number The physical coordinates corresponding to the column sensing area; Step S5-2: Process the coarse point cloud Statistical filtering is performed to remove outliers, resulting in an effective coarse point cloud. ; Step S5-3: Calculation The convex hull or minimum bounding rectangle on the horizontal plane is denoted as the boundary. ; Step S5-4: Set the boundary Extend outward by the preset margin distance Generate the boundary of the scanned region ,in The diameter is 5mm to 20mm; Step S5-5: Define the boundaries of the scanned area As the scanning field of view of the 3D vision camera, it drives the robot to move so that the range is within the field of view of the 3D vision camera.

10. The positioning method of a robot 3D vision positioning system based on area array TOF perception guidance according to claim 6, characterized in that, In step S6, when coordinating the adjustment of position and attitude: The control system is based on the local distance distribution matrix. ,in, Let be the distance value of the sensing area in the i-th row and j-th column, i=1-m, j=1-n, where m and n are the row and column numbers, respectively. Calculate the average distance μ, i.e.: ； The local normal vector is calculated by fitting the local plane using the least squares method. for: ； Let v be the direction vector of the optical axis of the 3D vision camera, then the included angle θ is: ； like Then, control the robot to rotate around the X and Y axes of the tool coordinate system, with the rotation angle increment being... ,in, This is a proportionality coefficient, with a value ranging from 0.1 to 0.

5. If μ exceeds the distance range, the robot moves along the optical axis, with a step size of [missing value]. ,in, This is the distance adjustment factor, with a value ranging from 0.2 to 0.8.

Images