An automatic docking method for irregular curved surface joint based on visual and force feedback

Abstract

The application discloses an automatic docking method for irregular curved surface joints based on visual and force feedback, and comprises the following steps: acquiring the spatial pose information of a male head through a three-dimensional camera; controlling a mechanical arm to carry a female head to an initial alignment position near the male head according to the spatial pose information; performing a small range search motion near the initial alignment position, and simultaneously collecting force data Fx, Fy, Fz, Mx, My and Mz in real time through a six-dimensional force sensor; at an optimal alignment point, controlling the mechanical arm to move downward along the Z direction and start rotating, so that the female head enters the clamping groove structure of the male head; in the process of rotating in, the rotating direction is dynamically adjusted; when it is detected that the direction of Fz changes or mutates, it is judged that the female head enters the locking structure area; and the docking is judged to be completed through the stable state of the force in the Z direction or the rotation angle reaching a preset value. The application realizes the automatic docking of irregular curved surface joints, improves the alignment accuracy and docking reliability, and reduces the safety risk of manual operation.

Description

Technical Field

[0001] This invention relates to the field of industrial automation assembly technology, specifically to general image data processing or generation, and more particularly to an automatic docking method for irregular curved surface joints based on vision and force feedback. Background Technology

[0002] The docking of liquid phase connectors for industrial fuels is typically done manually. However, these connectors are generally heavy, and there is a risk of liquid splashing during the docking process, posing a safety hazard. Existing automation solutions mostly rely on fixed trajectories or simple vision positioning, which are difficult to adapt to the precise docking requirements of non-standard curved surface connectors.

[0003] In recent years, automatic docking technology based on vision and force feedback has been extensively studied. For example, patent CN120438983A discloses a six-degree-of-freedom compliant adaptive integrated automatic docking method for large connectors based on vision and force feedback. This method uses a vision system to identify the pose information of the docking target and combines it with force sensor feedback for active compliant control, achieving automatic docking of large connectors. Furthermore, patent CN210605454U discloses a curved surface force control contour following device based on an XYZ axis slide table. This device uses a six-dimensional force / torque sensor and an industrial camera to form a detection feedback section, achieving contact force control between the end probe and the curved surface. Patent CN108542469B discloses a six-dimensional force sensor based on image feedback, which calculates the contact force by identifying the deformation of an elastic deformable body through image information.

[0004] However, the aforementioned existing technologies still have the following shortcomings: (1) Existing automatic docking methods mostly rely on motion planning after visual recognition. When the initial pose uncertainty of the joint is large, it is difficult to guarantee the alignment accuracy. (2) The existing technology does not involve an adaptive path adjustment mechanism during the screwing process. For irregular curved surface joints, the force situation during the screwing process is complex, and the fixed trajectory is prone to jamming or damage. (3) Existing technologies lack effective means for judging the locking area and locking control, making it difficult to achieve complete docking-locking closed-loop control.

[0005] Therefore, there is an urgent need for an automatic docking method that can adapt to irregular curved surface joints, has adaptive screw-in capability, and lock-in judgment function. Summary of the Invention

[0006] The purpose of this invention is to overcome the defects in the existing technology and provide an automatic docking method for irregular curved surface joints based on vision and force feedback. This method combines three-dimensional vision positioning with six-dimensional force feedback to achieve stable and high-precision automatic docking of irregular curved surface joints.

[0007] To achieve the above objectives, the technical solution of this invention is to design an automatic docking method for irregular curved surface joints based on vision and force feedback, comprising the following steps: S1: Obtain the spatial pose information of the male head through a 3D camera. The spatial pose information includes 3D position information and attitude information. S2: Based on the spatial pose information, control the robotic arm to move the female head to the initial alignment position near the male head, so that the axis of the female head is basically aligned with the axis of the male head and maintains a preset safety gap. S3: Perform a small-range search movement near the initial alignment position, and at the same time collect force data Fx, Fy, Fz, Mx, My, Mz in real time through a six-dimensional force sensor. By comparing the force at different positions, determine the position with the minimum force as the optimal alignment point, and record the height position H1 at this time. S4: At the optimal alignment point, control the robotic arm to move downward along the Z direction and start rotating, so that the female head enters the slot structure of the male head; S5: During the screwing process, the Z-direction force Fz data of the six-dimensional force sensor is collected in real time. By comparing the changing trend of Fz at continuous moments, the rotation direction is dynamically adjusted to keep Fz at its minimum or in a decreasing trend, thereby realizing an adaptive screwing path. S6: As the screw-in depth increases, when the difference between the current height H2 and H1 reaches the predetermined height, it is determined that the current area is near the locking region. Further monitoring of Fz mutation is conducted. When a change in direction or mutation of Fz is detected, it is determined that the female head has entered the locking structure region. S7: After entering the locking area, control the robotic arm to rotate in the opposite direction, so that the female head moves along the locking groove until it reaches the locking position; S8: The docking is considered complete when the Z-axis force stabilizes or the rotation angle reaches the preset value.

[0008] Fx, Fy, and Fz are the forces acting on the female connector in the X, Y, and Z directions, respectively, while Mx, My, and Mz are the torques acting on the female connector around the X, Y, and Z axes. The search motion in step S3 is performed after step S2, when the female and male connectors maintain a small gap (e.g., 1-3 mm) and may have already made slight contact. The essence of the search is to find the optimal alignment posture that minimizes forces in each direction while in contact. In steps S6 and S7, when the Z-direction force Fz exceeds the threshold, it is determined that the female connector has approached the locking structure area but has not entered correctly. At this point, the robotic arm is controlled to perform a short-stroke reverse rotation. The purpose of this rotation is to release the axial clamping force caused by misalignment and to guide the female connector relative to the male connector to the correct starting position of the locking structure using the connector's own guiding structure, such as a threaded chamfer or a slotted bevel. After completing the reverse rotation, the system again controls the robotic arm to rotate in the correct screw-in direction (which can be the same as or opposite to the direction in step S4, depending on the connector's locking direction), causing the female connector to move along the locking slot until it reaches the mechanical locking position.

[0009] A further technical solution is that, in step S3, the micro-range search motion is either a planar grid search or a spiral search.

[0010] A further technical solution is that, in step S5, the specific method of dynamically adjusting the rotation direction is as follows: when Fz continues to increase, the rotation direction is automatically reversed; when Fz decreases, the current rotation direction is maintained.

[0011] A further technical solution is that, in step S6, the predetermined height is a depth threshold pre-stored according to the male connector slot structure.

[0012] A further technical solution is that, in step S8, the stable state of the Z-direction force is that the fluctuation range of Fz is within a preset threshold and lasts for a preset time.

[0013] An automatic docking system for irregular curved surface joints based on vision and force feedback includes a three-dimensional camera for acquiring the spatial pose information of the male connector, a six-dimensional force sensor installed at the end of a robotic arm, a robotic arm for carrying the female connector to perform alignment, screwing in and locking actions, and a controller. The controller is connected to the three-dimensional camera, the six-dimensional force sensor and the robotic arm respectively. The controller includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements the steps of the automatic docking method for irregular curved surface joints based on vision and force feedback as described in any one of claims 1 to 5.

[0014] The six-dimensional force sensor is used to collect force data in real time during the docking process between the female and male connectors.

[0015] A further technical solution is to use a structured light camera or a binocular vision camera as the 3D camera.

[0016] A further technical solution is that the six-dimensional force sensor is a strain gauge or capacitive six-dimensional force sensor.

[0017] The advantages and beneficial effects of this invention are as follows: 1. Enable automatic docking of irregular curved surface joints; 2. Improve alignment accuracy through a minimum force search mechanism; 3. Adaptive adjustment of the screw-in path is achieved through changes in Z-axis force; 4. By judging the locking state through force changes, the reliability of docking is improved; 5. Reduce reliance on manual labor and improve operational safety and efficiency. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a first embodiment of the automatic docking system for irregular curved surface joints based on vision and force feedback according to the present invention; Figure 2 yes Figure 1 A magnified view of a portion of the headpiece; Figure 3 yes Figure 2 A picture of the actual product; Figure 4 This is a schematic diagram of Embodiment 5 of the present invention; Figure 5 yes Figure 4 A magnified view of the central elliptical portion.

[0019] In the diagram: 1. 3D camera; 2. Robotic arm; 3. Six-dimensional force sensor; 4. Female connector; 5. Male connector; 6. Card slot; 7. Tank; 8. Base body; 9. Elastic element. Detailed Implementation

[0020] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and examples. The following examples are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0021] In this invention, the robotic arm end effector has at least two operating modes: a force-compliant mode and a position-controlled mode. In the force-compliant mode, the robotic arm end effector actively adjusts its position and posture based on feedback data from a six-dimensional force sensor to reduce contact force. In the position-controlled mode, the robotic arm end effector moves along a predetermined trajectory, maintaining a rigid connection and not actively yielding due to contact force.

[0022] Example 1: This invention is an automatic docking method for irregular curved surface joints based on vision and force feedback. The specific steps are as follows: S1: Acquire the spatial pose information of the male head mounted on the tank using a 3D camera. The 3D camera can be a structured light camera, mounted on a fixed bracket outside the robotic arm, to acquire the 3D position (x, y, z) and orientation (Rx, Ry, Rz) of the male head.

[0023] S2: Based on the visual positioning results, the controller controls the robotic arm to move the female head to the initial alignment position near the male head, so that the axis of the female head is basically aligned with the axis of the male head, and maintains a safety gap of about 5mm.

[0024] S3: Perform a small-range planar mesh search near the initial alignment position with a step size of 0.5 mm, while simultaneously acquiring Fx, Fy, Fz, Mx, My, and Mz data in real time using a six-dimensional force sensor. The computer compares the combined force amplitude sqrt(Fx) at each search point. 2 +Fy 2 +Fz 2 Determine the position with the least force as the optimal alignment point and record the height H1 at this point.

[0025] S4: At the optimal alignment point, control the robotic arm to move downwards along the Z direction and begin rotating forward at a speed of 5 rpm, so that the female head enters the slot 6 structure of the male head 5 (e.g., Figure 2 , Figure 3 (As shown).

[0026] S5: During the screw-in process, the Z-axis force Fz is monitored in real time. When Fz is greater than the previous value for three consecutive samples, the rotation direction is automatically reversed; when Fz decreases, the current direction is maintained. This achieves adaptive screw-in. The robotic arm end effector defaults to force-controlled compliant mode, dynamically adjusting the rotation direction according to the changing trend of Fz.

[0027] S6: When H2-H1 reaches the preset depth threshold of 8mm, it is determined that the device has entered the vicinity of the locking area. At this time, Fz is continuously monitored. If Fz suddenly changes by more than 30% from a stable value or changes from positive to negative, it is determined that the female head has entered the locking structure area.

[0028] S7: After entering the locking area, control the robotic arm to rotate in the opposite direction, so that the female head moves along the locking groove until the Z-axis force stabilizes near the minimum value.

[0029] S8: When the Z-axis force fluctuation is less than ±1N and lasts for 0.5 seconds, or when the reverse rotation angle reaches the preset value of 60°, the docking is considered complete.

[0030] Example 2 This embodiment provides an automatic docking system for irregular curved surface joints based on vision and force feedback, the structure of which is as follows: Figure 1As shown, the system includes: a 3D camera 1 for acquiring 3D point cloud data of the male head 5; a strain-type six-dimensional force sensor 3 installed between the end flange of the robotic arm 2 and the female head gripper; a six-axis industrial robotic arm for carrying the female head 4 to perform motion; and a controller. The 3D camera is fixedly installed above the work area; the sampling frequency of the six-dimensional force sensor is 1000Hz; the controller uses an industrial PC to run a real-time control system and execute the method described in Example 1. The 3D camera can be a structured light camera. The six-dimensional force sensor is installed between the end flange of the robotic arm and the gripper for holding the female head. When the robotic arm drives the female head to move and makes physical contact with the male head, the six-dimensional force sensor detects in real time the force signals Fx, Fy, and Fz of the female head in three orthogonal directions in the Cartesian coordinate system, as well as the torque signals Mx, My, and Mz around these three directions, and feeds these signals back to the robot's control system in real time.

[0031] The controller receives point cloud data from the 3D camera 1 and calculates the pose of the male head 5 using a point cloud registration algorithm; it receives force / torque data from the six-dimensional force sensor 3 and performs force control algorithm calculations; it sends motion commands to the robotic arm 2 to achieve hybrid position servo and force control.

[0032] The rest is the same as in Example 1.

[0033] Example 3: In actual industrial scenarios, fuel tanks may experience local surface deformation such as dents or bulges after long-term use, leading to a significant deviation between the actual installation posture of the male connector and the design value. To address this technical challenge, this example provides an adaptive surface compensation docking method based on 3D vision guidance. The difference from Example 1 is that, before acquiring the spatial pose information of the male connector in step S1, the following procedure is included: controlling a 3D camera to perform multi-view scanning of the male connector installation area on the tank along a preset trajectory, acquiring at least three local point cloud data images from different angles. The controller registers and fuses the point cloud data from each viewpoint to reconstruct a local surface 3D model of the male connector installation area.

[0034] Specifically, the iterative nearest-point algorithm is used to rigidly register the point clouds of adjacent viewpoints, minimizing the registration error function: E(R,t)= ||pi′−(Rpi+t)|| 2 Where pi and pi′ are corresponding point pairs in the point clouds of adjacent viewpoints, R is the rotation matrix, and t is the translation vector. After registration, a dense 3D surface model of the installation area is generated.

[0035] Before step S2, the following steps are also included: comparing the reconstructed local surface 3D model with the pre-stored standard surface model, and calculating the surface deformation characteristics of the installation area, specifically: the surface normal deviation angle θ. n Local curvature change Δκ, deformation gradient direction dd → ;wherein the surface normal deviation angle θ n It is the angle between the actual surface normal and the design normal; the local curvature change Δκ is the difference between the actual curvature and the design curvature; the deformation gradient direction d d → It is the direction of maximum change in surface deformation.

[0036] Based on the surface deformation characteristics, adaptive compensation is performed on the male head spatial pose information obtained in step S1. The compensated male head pose T comp Represented as: T comp =T vision .ΔT(θ n ,Δκ,d d → Among them, T vision The nominal pose of the male head is obtained directly from the 3D camera measurement, and ΔT is the deformation compensation transformation matrix. Where T... comp The orientation component in the equation is the compensation direction, which is used to guide the subsequent movement path of the robotic arm.

[0037] The robotic arm is controlled to move the female head to the compensated initial alignment position, ensuring that the female head axis is substantially aligned with the compensated male head axis. The compensated docking path effectively avoids docking interference caused by deformation of the tank surface.

[0038] The subsequent precise alignment search, adaptive screw-in and locking judgment steps are basically the same as steps S3 to S8 in Embodiment 1. The difference is that during the adaptive screw-in process, the initial direction of the screw-in path adopts the orientation component in the compensated male head pose.

[0039] Example 4: The difference between this example and Example 1 lies in the screw-in control method in step S5. In traditional docking methods, when increased screw-in resistance is detected, the control system typically reduces the rotation speed, increases flexibility, or passively yields to avoid damaging the joint. However, this example finds that for slots with helical inclined surfaces (such as threads, helical slots, etc.), the flexible yielding strategy can actually prevent the female head from sliding into the correct slot along the inclined surface. The core innovation of this example is to do the opposite, utilizing the rigid connection at the end of the robotic arm to force the female head to slide in along the helical inclined surface when encountering screw-in resistance.

[0040] In step S5, when Fz and Mz are detected to increase simultaneously, the control system switches the end effector of the robotic arm from force control compliant mode to position control mode, maintaining a rigid connection state without reducing the speed or yielding, and using the spiral ramp for forced guidance.

[0041] The specific steps are as follows: After the screwing begins in step S4, the control system switches the stiffness parameter of the robotic arm's end effector to a high-stiffness mode. Specifically, the controller disables the force-controlled compliant mode and adopts a position-controlled primary control strategy to ensure that the robotic arm's end effector maintains a rigid connection state when subjected to axial and tangential forces, without producing active yielding or passive deflection.

[0042] The Z-direction force Fz and torque Mz of the six-dimensional force sensor are acquired in real time, and the rate of change of Fz ΔFz / Δθ within a unit rotation angle is calculated. When ΔFz / Δθ exceeds a preset threshold, it is determined that the female thread teeth or jaws are colliding with the edge of the male slot, and the forced guidance stage is initiated.

[0043] Upon detecting a collision, the control system neither reduces the rotational speed nor yields; instead, it maintains or slightly increases the axial feed rate. Utilizing the geometric constraint of the helical ramp between the female and male connector slots, the rotational motion is forcibly converted into axial feed motion. The specific mechanical principle is as follows: When the female thread crest is on the edge of the male slot, the contact force can be decomposed into a normal component Fn and a tangential component Ft. Due to the presence of the helical ramp, the normal component Fn generates an axial component Fz_guide = Fn.sinα (α is the helix angle), which pushes the female thread to slide along the ramp. Under rigid connection conditions, the rotational torque continuously applied by the robotic arm maintains the normal component Fn, ensuring that the female thread is continuously subjected to an axial guiding force until the thread slides past the edge and into the correct slot.

[0044] When Fz suddenly drops and Mz returns to a stable state, it is determined that the female head has successfully entered the correct slot. At this point, the normal screw-in speed is resumed to continue docking.

[0045] This embodiment eliminates the need for visual recognition of the thread start phase, utilizing the geometric characteristics of the mechanical structure itself to achieve automatic "thread finding"; it avoids the "slippage" problem caused by flexible yielding, thus improving the success rate of screwing in; compared with the existing technology that adds a flexible mechanism, it requires no additional hardware and can be achieved simply by adjusting the control strategy.

[0046] Existing technology reveals that during the docking process of irregular curved surface joints, even with visual positioning and force feedback control, slight attitude deviations may still exist between the robotic arm's end effector and the male connector. Because traditional robotic arm end effects are rigid connections, when there is an angular deviation between the female and male connectors, the rigid connection can lead to localized stress concentration, potentially damaging the joint structure or causing docking failure. The consequences of this problem include: joint jamming or jamming; scratches or damage to the joint surface; and the need for manual intervention in docking failures.

[0047] Furthermore, when the female connector begins to rotate into the male connector's groove, due to surface deformation or visual positioning errors, the threads or jaws of the female connector may collide with the edge of the male connector's groove instead of sliding smoothly into it. While existing technologies employ curved surface mating structures to improve torsional resistance and sealing, they still fall short in addressing the "thread finding" issue during the initial docking stage. This problem results in increased screwing resistance triggering false alarms, wear on the connector threads / grooves, and prolonged docking time.

[0048] This embodiment overturns the traditional understanding of "flexible docking." Utilizing the rigid connection at the end of the robotic arm and the geometric self-guiding characteristics of the helical ramp, it forces the female end to slide into the correct slot when encountering screwing resistance. Compared to existing technologies that add flexible mechanisms or reduce rotational speed, this embodiment requires no additional hardware and avoids the "slippage" problem, increasing the screwing success rate by over 30%. This embodiment employs reverse thinking, using end-effector rigidity to solve the screwing deviation problem.

[0049] Example 5: The difference between this example and Example 1 lies in the installation method of the male connector. In traditional docking solutions, the male connector is usually rigidly fixed to the tank to ensure the stability of its position during docking. However, this example reveals that this rigid fixing method can cause the female connector threads to collide with the male connector's groove edge when there is a slight phase deviation between the female and male connectors, increasing docking resistance.

[0050] The core innovation of this embodiment is to do the opposite and make the male connector "move"—the male connector is installed through an elastic base, so that it can be passively offset when it is subjected to the contact force of the female connector, actively "giving way" to help the female connector enter the correct slot.

[0051] like Figure 4 , Figure 5 As shown, the specific structure and working steps are as follows: Before docking begins, the male connector is secured to the tank using the flexible mounting base.

[0052] The flexible mounting base includes a base body 8 fixed to the surface of the tank 7, an elastic element 9 disposed between the base body and the male connector, and a reset mechanism that automatically returns the male connector to its initial position after docking.

[0053] The base body 8 has a cavity for accommodating the male connector; the elastic element 9 is a polyurethane rubber pad with nonlinear stiffness characteristics, which provides low elastic stiffness in the small deformation range (0~1mm) to allow the male connector to passively deflect, and the stiffness increases sharply in the large deformation range (>1mm) to limit the excessive deflection of the male connector; the reset mechanism can be achieved by using a pre-compressed spring or the self-rebound force of the elastic element 9.

[0054] The flexible mounting base allows the male connector to have a passive range of motion of ±1~2mm in the radial and circumferential directions, and a passive range of motion of ±0.5~1mm in the axial direction.

[0055] During the screwing-in process in step S5, when the female thread or jaws collide with the edge of the male slot, the male head passively shifts on the elastic mounting base under the action of contact force. Specifically, when there is radial eccentricity between the female and male heads, the male head shifts laterally under the action of radial contact force, allowing the female head to align with the slot entrance; when the female thread collide with the edge of the male slot, the male head rotates slightly under the action of tangential force, causing the edge to "give way," allowing the female head to slide into the slot; and when the axial thrust of the female head is too large, the male head can passively retract to avoid hard impact.

[0056] After the female connector enters the correct slot, the male connector automatically returns to its centered position under the action of elastic restoring force. During subsequent screwing, the male connector continues to passively follow the female connector, automatically fine-tuning its position according to the movement of the female connector to keep the screwing resistance at a minimum.

[0057] After docking is completed, the male connector automatically returns to its initial position under the action of the reset mechanism of the elastic base, preparing for the next docking.

[0058] This embodiment breaks the traditional misconception that "the male head must be rigidly fixed," and utilizes the passive mobility of the male head to assist in docking. Compared with the flexible solution on the female side, the elastic yielding on the male side does not affect the control accuracy of the robotic arm, and has a simple structure and low cost. The yielding on the male side and the active control on the female side form a "two-way synergy," which significantly improves the docking success rate. The elastic base also has a vibration isolation function, which can reduce the interference of high-frequency vibration on the force sensor data.

[0059] As is known from existing technology, male connectors are typically mounted on irregular curved surfaces such as tanks, and the curvature of the mounting surface affects the actual orientation and stability of the male connector. In existing technologies, male connector mounting bases are mostly planar structures, which may lead to problems such as loose fit and uneven stress when connected to curved tank surfaces. The impact of this problem includes: unstable male connector mounting posture, affecting docking accuracy; when the tank surface deforms, the male connector axis may deviate beyond the visual positioning range; and the male connector may loosen after long-term use.

[0060] Furthermore, existing technology reveals that when the female connector begins to rotate into the male connector's groove, due to surface deformation or visual positioning errors, the threads or jaws of the female connector may collide with the edge of the male connector's groove instead of sliding smoothly into it. While existing technologies employ curved surface mating structures to improve torsional resistance and sealing, they still fall short in addressing the "thread finding" issue during the initial docking stage. The consequences of this problem include: increased screwing resistance triggering false alarms; wear on the connector threads / grooves; and prolonged docking time.

[0061] This embodiment takes the opposite approach, breaking the traditional misconception that "the male connector must be rigidly fixed." It uses a flexible mounting base to allow the male connector to move passively. When the female connector collides with the male connector, the male connector actively "yields," assisting the female connector into the correct slot. Compared to the flexible female-side solution, this embodiment does not affect the robot arm's control accuracy and has a simpler structure and lower cost. This embodiment uses reverse thinking, utilizing the male connector's non-fitting mounting to solve the problem of screw-in deviation.

[0062] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An automatic docking method for irregular curved surface joints based on vision and force feedback, characterized in that, Includes the following steps: S1: Obtain the spatial pose information of the male head through a 3D camera. The spatial pose information includes 3D position information and attitude information. S2: Based on the spatial pose information, control the robotic arm to move the female head to the initial alignment position near the male head, so that the axis of the female head is basically aligned with the axis of the male head and maintains a preset safety gap. S3: Perform a small-range search movement near the initial alignment position, and at the same time collect force data Fx, Fy, Fz, Mx, My, Mz in real time through a six-dimensional force sensor. By comparing the force at different positions, determine the position with the minimum force as the optimal alignment point, and record the height position H1 at this time. S4: At the optimal alignment point, control the robotic arm to move downward along the Z direction and start rotating, so that the female head enters the slot structure of the male head; S5: During the screwing process, the Z-direction force Fz data of the six-dimensional force sensor is collected in real time. By comparing the changing trend of Fz at continuous moments, the rotation direction is dynamically adjusted to keep Fz at its minimum or in a decreasing trend, thereby realizing an adaptive screwing path. S6: As the screw-in depth increases, when the difference between the current height H2 and H1 reaches the predetermined height, it is determined that the current area is near the locking region. Further monitoring of Fz mutation is conducted. When a change in direction or mutation of Fz is detected, it is determined that the female head has entered the locking structure region. S7: After entering the locking area, control the robotic arm to rotate in the opposite direction, so that the female head moves along the locking groove until it reaches the locking position; S8: The docking is considered complete when the Z-axis force stabilizes or the rotation angle reaches the preset value.

2. The automatic docking method for irregular curved surface joints based on vision and force feedback according to claim 1, characterized in that, In step S3, the small-range search motion is either a planar grid search or a spiral search.

3. The automatic docking method for irregular curved surface joints based on vision and force feedback according to claim 2, characterized in that, In step S5, the specific method for dynamically adjusting the rotation direction is as follows: when Fz continues to increase, the rotation direction is automatically reversed; when Fz decreases, the current rotation direction is maintained.

4. The automatic docking method for irregular curved surface joints based on vision and force feedback according to claim 3, characterized in that, In step S6, the predetermined height is a depth threshold pre-stored according to the male connector slot structure.

5. The automatic docking method for irregular curved surface joints based on vision and force feedback according to claim 4, characterized in that, In step S8, the stable state of the Z-direction force is that the fluctuation range of Fz is within a preset threshold and lasts for a preset time.

6. An automatic docking system for irregular curved surface joints based on vision and force feedback, characterized in that, It includes a 3D camera for acquiring the spatial pose information of the male head, a six-dimensional force sensor mounted on the end of the robotic arm, a robotic arm for carrying the female head to perform alignment, screwing and locking actions, and a controller. The controller is connected to the three-dimensional camera, the six-dimensional force sensor and the robotic arm respectively. The controller includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it implements the steps of the automatic docking method for irregular curved surface joints based on vision and force feedback as described in any one of claims 1 to 5.

7. The automatic docking system for irregular curved surface joints based on vision and force feedback according to claim 6, characterized in that, The 3D camera is a structured light camera or a binocular vision camera.

8. The automatic docking system for irregular curved surface joints based on vision and force feedback according to claim 6, characterized in that, The six-dimensional force sensor is either a strain gauge type or a capacitive six-dimensional force sensor.

Images