Fully automatic production embodied intelligent robot self-adapting assembly method and system

Abstract

The application provides a full-automatic production body-integrated intelligent robot adaptive assembly method and system, relates to the technical field of industrial robot assembly, and acquires the limiting direction, contact state and track-changing impact direction of a positioning part when a four-way shuttle vehicle changes tracks and transports a tray, determines a main positioning group, a reference positioning group and a to-be-inspected positioning part that is separated from an end face after track changing; locks the tray and keeps the main positioning face and the lateral reference face of the to-be-assembled body in abutment, controls the body-integrated intelligent robot to perform reverse repeated abutment, forward separation and reverse verification along the intersection line direction of the two positioning faces; according to bidirectional detection data, calibrated contact poses and end poses at each contact time, distinguishes the body off-position, positioning slip and invalid detection state, and determines an assembly reference pose; according to the assembly reference pose and the lateral hole position, locally corrects the reversing wheel assembly track, and inserts the reversing wheel positioning shaft into the lateral mounting hole of the to-be-assembled body. The application can be used for adaptive correction of the assembly track after track-changing impact.

Description

Technical Field

[0001] This application relates to the field of industrial robot assembly technology, and in particular to an adaptive assembly method and system for fully automated production using an embodied intelligent robot. Background Technology

[0002] On automated production lines, mobile tooling pallets can be transported by a four-way shuttle between longitudinal and transverse tracks to sequentially deliver thin-walled box-shaped components to inspection and assembly stations. To balance repetitive positioning and workpiece handling, pallets typically employ a combination of bottom fixed support, lateral positioning, and end-retractable positioning, rather than rigidly fixing the components to be assembled without release.

[0003] Existing fully automated assembly methods typically read the pallet markings after the pallet arrives at the assembly station and call up a fixed assembly trajectory according to the pre-calibrated pallet coordinate system. Some production lines also use vision cameras to detect lateral mounting holes and translate the assembly trajectory as a whole based on the detection results, or use force sensors at the end of the robot to perform compliance compensation during the insertion process.

[0004] However, during the four-way shuttle track changing, the reversing wheel switching the bearing direction, the lifting mechanism's operation, and the track joints will create transient impacts within the pallet plane. When the bottom support block and the side positioning block remain in contact, but the lateral retractable positioning component disengages from the end face of the body to be assembled, the seemingly identical disengagement state may be caused by two different reasons: firstly, the body to be assembled may have moved along the lateral direction while the lateral positioning component has not moved; secondly, the lateral positioning component may have slid relative to the pallet, causing the body to be assembled to shift along with the lateral positioning reference after the slippage.

[0005] In the first scenario, simply repositioning the main body to be assembled to its original calibration position allows the initial assembly trajectory to be reused. In the second scenario, if the original calibration position is still used as the reference, a systematic misalignment will occur between the reversing wheel positioning shaft and the lateral mounting hole. A single contact is insufficient to distinguish between the dislocation of the main body and the slippage of the positioning component; directly correcting the trajectory based on the contact position may introduce abrupt changes in trajectory when the reversing wheel passes through the narrow channel formed by the reference positioning group. Furthermore, if visual inspection is used, the lateral mounting hole may be affected by fixture obstruction, housing reflections, and hole depth, ultimately leading to assembly misalignment. Summary of the Invention

[0006] The purpose of this application is to provide an adaptive assembly method and system for fully automated production using an embodied intelligent robot, in order to solve the problem in the prior art that it is difficult to distinguish between the main body displacement and the slippage of the positioning parts in the scenario of track changing impact of a four-way shuttle, which leads to assembly misalignment.

[0007] To address the aforementioned technical problems, in a first aspect, this application provides an adaptive assembly method for a fully automated production system using an embodied intelligent robot, the method comprising: During the process of transporting the pallet by the four-way shuttle, the limiting direction of each positioning component of the pallet, the contact state with the main body to be assembled, the impact direction of the four-way shuttle, the calibrated contact posture, the lateral hole position of the main body to be assembled on the pallet, and the initial assembly trajectory are obtained. Based on the restricted direction, contact state, and track-changing impact direction, determine the main positioning group, the reference positioning group, and the positioning component to be inspected that detaches from the end face after track changing; The tray is locked so that the main positioning group and the reference positioning group respectively abut against the main positioning surface and the lateral reference surface of the receiving assembly body; During the contact process, the embodied intelligent robot holds the main body to be assembled and sequentially contacts, separates from, and contacts the positioning part to be inspected in the opposite direction along the intersection of the main positioning surface and the lateral reference surface, recording the contact sequence and detection force to form bidirectional detection data. Based on the bidirectional detection data, the calibrated contact pose, and the end pose of the embodied intelligent robot during reverse contact, forward disengagement, and re-reverse contact, the positioning state and assembly reference pose of the positioning component to be inspected are determined. Based on the positioning state, assembly reference pose, and lateral hole position, the assembly trajectory of the reversing wheel, which is the part to be assembled, is determined, and the reversing wheel is inserted into the lateral mounting hole of the main body to be assembled along the assembly trajectory.

[0008] Optionally, the embodied intelligent robot clamps the body to be assembled and, along the intersection of the main positioning surface and the lateral reference surface, sequentially contacts, disengages, and re-contacts the positioning component to be inspected in the reverse direction, records the contact sequence and detection force, and forms bidirectional detection data, including: Based on the normal of the main positioning surface, the normal of the lateral reference surface, the limiting direction of the positioning component to be inspected, and the impact direction of the track changing, determine the direction of the intersection line with positive and negative directions; With both the main positioning group and the reference positioning group in contact, the main body to be assembled is moved along the intersection direction, and the moment of contact change of the positioning part to be inspected is determined as the segmentation moment. According to the segmented time, the movement process of the main body to be assembled is divided into a reverse reconnection segment, a forward disengagement segment, and a reverse verification segment. The contact sequence, detection force, end displacement, and end pose of the embodied intelligent robot corresponding to the reverse reconnection segment, forward disengagement segment, and reverse verification segment are arranged according to the sampling time to form bidirectional detection data.

[0009] Optionally, determining the intersection direction with positive and negative directions based on the normal of the main positioning surface, the normal of the lateral reference surface, the limiting direction of the positioning component to be inspected, and the impact direction of the track changing includes: By cross-product of the normal of the main positioning surface and the normal of the lateral reference surface, the contact tangent of the body to be assembled is obtained when the main positioning group and the reference positioning group are in contact. Projecting the limiting direction of the positioning component under test onto the contact tangent, we obtain the candidate direction; When the projection of the track-changing impact direction onto the candidate direction is positive, the candidate direction is determined as positive, and the opposite direction of the candidate direction is determined as negative, thus forming the intersection direction.

[0010] Optionally, the step of dividing the movement process of the main body to be assembled into a reverse reconnection segment, a forward disengagement segment, and a reverse verification segment according to the segmented time includes: The smaller value between the retraction stroke of the positioning component to be inspected and the non-interference stroke of the embodied intelligent robot is determined as the detection stroke, and the upper limit of the force on the end face of the main body to be assembled is determined as the upper limit of the detection force. At the beginning of each detection segment, the end displacement is re-accumulated, and the detection force along the intersection direction and the contact status of the reference positioning group are monitored. Under the conditions that the end displacement is less than the detection stroke, the detection force is less than the upper limit of the detection force, and the contact state is that the reference positioning group is in contact, the body to be assembled is moved in the opposite direction of the intersection line until the positioning part to be inspected contacts, forming a reverse reconnection section. Move the main body to be assembled from the end of the reverse reconnection section along the positive direction of the intersection line until the positioning part to be inspected is disengaged, forming the positive disengagement section; From the end of the forward separation section, move the assembly body again in the opposite direction of the intersection line until it contacts the positioning part to be inspected, forming the reverse verification section.

[0011] Optionally, determining the positioning state and assembly reference pose of the component to be inspected based on the bidirectional detection data, the calibrated contact pose, and the end-effector pose of the embodied intelligent robot during reverse contact, forward disengagement, and re-reverse contact includes: The detection force at the first contact, the detection force at the separation, and the detection force at the second contact are extracted from the bidirectional detection data, and associated with the corresponding end poses to form reconnection data. Based on the reconnection data, determine the reconnection difference between the two contact positions, the force difference between the two contact forces, and the loop difference enclosed by the contact and disengagement processes; The positional difference between the end pose at the second contact and the calibrated contact pose is determined as the reference difference; The aforementioned connection difference, force difference, hysteresis difference, and reference difference constitute the positioning features; The positioning state of the positioning component to be inspected is determined based on the positioning features, and the assembly reference pose is selected from the calibrated contact pose and the end pose at the second contact based on the positioning state.

[0012] Optionally, the step of determining the positioning state of the positioning component to be inspected based on the positioning features, and selecting the assembly reference pose from the calibrated contact pose and the end pose at the second contact based on the positioning state, includes: When the reconnection difference is not greater than the reconnection threshold, the force difference is not greater than the force difference threshold, the return line difference is not greater than the return line threshold, and the reference difference is not greater than the preset reference threshold, the positioning state is determined as the main body is off-position, and the calibrated contact posture is determined as the assembly reference posture. When the reconnection difference is not greater than the reconnection threshold, the force difference is not greater than the force difference threshold, the return line difference is not greater than the return line threshold, and the reference difference is greater than the preset reference threshold, the positioning state is determined as positioning slip, and the end pose at the second contact is determined as the assembly reference pose. When the repeat difference is greater than the repeat threshold, the force difference is greater than the force difference threshold, or the lap difference is greater than the lap threshold, the positioning status is determined as invalid detection, and the determination of the assembly reference pose is stopped.

[0013] Optionally, determining the assembly trajectory of the reversing wheel, which is the part to be assembled, based on the positioning state, assembly reference pose, and lateral hole position includes: Calculate the positional difference between the assembly reference pose and the calibration contact pose in the limiting direction of the positioning part to be inspected, and obtain the candidate correction amount; When the positioning state is that the main body is out of position, zero is determined as the hole position correction amount; when the positioning state is that the positioning is slipping, the candidate correction amount is determined as the hole position correction amount. Based on the position of the reference positioning group and the lateral hole position, the correction start point and correction end point are determined on the initial assembly trajectory, and the trajectory segment between the correction start point and the correction end point is determined as the correction range; Based on the position of each trajectory point within the correction range, each trajectory point is translated and corrected along the limiting direction of the positioning component to be inspected, and the initial assembly trajectory outside the correction range is retained to obtain the assembly trajectory of the reversing wheel.

[0014] Optionally, the step of determining the correction start point and correction end point on the initial assembly trajectory based on the position of the reference positioning group and the lateral hole position, and defining the trajectory segment between the correction start point and the correction end point as the correction range, includes: By referring to the material-side edge points of the two positioning components in the positioning group, an inlet surface perpendicular to the main positioning surface is constructed; Create an opening surface perpendicular to the axis of the lateral mounting hole by passing through the center of the opening; The intersections of the initial assembly trajectory with the inlet surface and the orifice surface are respectively determined as the correction start point and correction end point, and the initial assembly trajectory between the correction start point and the correction end point is determined as the correction range.

[0015] Optionally, inserting the reversing wheel into the lateral mounting hole of the body to be assembled along the assembly trajectory includes: When the main positioning group and the reference positioning group are in contact and the positioning state is not invalid, the control station clamping component clamps the body to be assembled and keeps the body to be assembled in the assembly reference position. After detecting the clamping signal of the clamping component at the workstation, the intelligent robot is controlled to release the main body to be assembled and clamp the reversing wheel; Move the reversing wheel along the assembly trajectory to the reference positioning group so that the positioning axis of the reversing wheel corresponds to the lateral mounting hole, forming the entrance pose; The reversing wheel is moved along the normal of the lateral reference plane from the entrance position, and the force exerted by the reversing wheel along the limiting direction of the positioning component under test is restricted. When the positioning shaft of the reversing wheel enters the lateral mounting hole and the insertion depth reaches the predetermined depth, the reversing wheel stops moving.

[0016] Secondly, this application provides a fully automated production system for an embodied intelligent robot adaptive assembly system, the system comprising: The acquisition module is used to acquire the limiting direction of each positioning component of the pallet, the contact state with the main body to be assembled, the impact direction of the four-way shuttle changing track, the calibrated contact posture, the lateral hole position of the main body to be assembled on the pallet and the initial assembly trajectory during the process of the four-way shuttle changing track transport pallet. The first determining module is used to determine the main positioning group, the reference positioning group, and the positioning component to be inspected that detaches from the end face after the track change, based on the limiting direction, the contact state, and the track-changing impact direction. A locking module is used to lock the pallet so that the main positioning group and the reference positioning group respectively abut against the main positioning surface and the lateral reference surface of the receiving assembly body; The clamping module is used to clamp the body to be assembled by an embodied intelligent robot during the contact process, and sequentially contact the positioning part to be inspected in the reverse direction, separate it in the forward direction, and contact it again in the reverse direction along the intersection of the main positioning surface and the lateral reference surface, and record the contact sequence and detection force to form bidirectional detection data. The second determining module is used to determine the positioning state and assembly reference posture of the positioning component to be inspected based on the bidirectional detection data, the calibrated contact posture, and the end posture of the embodied intelligent robot when it makes reverse contact, forward separation, and reverse contact again. The insertion module is used to determine the assembly trajectory of the reversing wheel, which is the part to be assembled, based on the positioning state, assembly reference pose and lateral hole position, and insert the reversing wheel into the lateral mounting hole of the main body to be assembled along the assembly trajectory.

[0017] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute the computer program to implement the steps of the fully automated production adaptive assembly method for an embodied intelligent robot as described in the first aspect above.

[0018] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, enables the implementation of the steps of the fully automated production adaptive assembly method for an embodied intelligent robot as described in the first aspect above.

[0019] The fully automated production adaptive assembly method of the embodied intelligent robot provided in this application has the following beneficial effects: under the condition that the main positioning surface and the lateral reference surface are kept in contact, the embodied intelligent robot clamps the main body to be assembled and releases the degree of freedom along the end direction to perform three-stage detection, so that the detection displacement corresponds to the reference change in the restricted direction of the positioning part to be inspected; by utilizing the consistency of the two reconnections and the deviation of the second reconnection position relative to the calibration position, the main body displacement and the positioning part slippage are distinguished, avoiding the continued use of the end calibration reference that has already shifted.

[0020] This application only performs translational correction on the trajectory between the reversing wheel entering the reference positioning group and reaching the lateral mounting hole, and can set an incremental transition at the correction starting point to preserve the existing trajectory of the picking segment and the long-distance movement segment.

[0021] This application utilizes a station clamping component to maintain the assembly reference posture of the main body to be assembled during the gripper change of the embodied intelligent robot. After receiving the clamping signal from the station clamping component, the embodied intelligent robot releases the main body to be assembled and switches from the main body gripper to the reversing wheel gripper to clamp the reversing wheel and perform insertion. This ensures that the assembly reference posture determined by the detection is maintained throughout the insertion process, reducing assembly misalignment caused by the main body to be assembled leaving the position again during the gripper change. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating an adaptive assembly method for a fully automated production using an embodied intelligent robot, provided in an embodiment of this application. Figure 2 This is a structural schematic diagram of the tray, the body to be assembled, and the positioning component provided in an embodiment of this application; Figure 3A schematic diagram illustrating the reverse reconnection, forward disconnection, and reverse verification process provided in an embodiment of this application; Figure 4 This is a schematic diagram of assembly trajectory correction provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a fully automated production embodied intelligent robot adaptive assembly system provided in an embodiment of this application. Detailed Implementation

[0024] To address the issue that in existing fully automated production processes, the impact of a four-way shuttle changing track can cause the assembly subject on the pallet to detach from the endocentric positioning component. Existing methods, relying solely on a single contact position, struggle to distinguish between temporary detachment of the assembly subject and slippage of the endocentric positioning component relative to the pallet, leading to distorted assembly datum, inaccurate lateral hole calculations, and misalignment of the reversing wheel's insertion trajectory. This application provides an adaptive assembly method using an embodied intelligent robot for fully automated production. The core idea of ​​this method is to perform bidirectional reconnection detection on the detached positioning component after track changing, while maintaining contact between the main positioning group and the reference positioning group. This distinguishes between the detachment of the assembly subject and the slippage of the positioning component, and corrects the reversing wheel's assembly trajectory based on the assembly datum pose determined by the detection. Therefore, this application can identify the detachment of the subject and the slippage of the positioning component without releasing the bottom and side constraints of the assembly subject, avoiding the continued use of the distorted endocentric calibration datum, and ensuring that the reversing wheel's positioning shaft completes insertion according to a trajectory adapted to the current position of the lateral mounting hole.

[0025] To facilitate understanding of this application, some terms are explained as follows: A pallet is a mobile tooling base that is transported between longitudinal and transverse tracks by a four-way shuttle and locked at the assembly station.

[0026] Positioning components refer to support blocks, lateral positioning blocks, or end positioning blocks that are mounted on a pallet and form contact constraints with the body to be assembled through positioning surfaces. The data for each positioning component can be represented as a combination of structural identification, limiting direction, contact state, positioning surface position, and allowable clearance.

[0027] Among them, the structural identifier is used to associate entities in the pallet model, and the constraint direction is the direction in which the positioning component prevents the main body to be assembled from continuing to approach the positioning component. In data processing, it is represented as a unit vector in the pallet coordinate system.

[0028] The contact state is a quantity indicating whether an effective abutment is formed between the positioning surface of the positioning component and the corresponding surface of the body to be assembled. It can be expressed using two values: contact and disengagement. When the pressure value of the pressure detection component reaches the contact threshold and remains in contact for the specified duration, the state is set to contact; when the pressure value decreases to the release threshold and remains in release for the specified duration, the state is set to disengagement, where the release threshold is less than the contact threshold. By setting different contact and release thresholds, frequent switching of the contact state can be avoided when the pressure value fluctuates around a single threshold due to track switching vibration.

[0029] The positioning surface position is used to determine the nominal contact point between the positioning element and the body to be assembled, and the allowable clearance is used to limit the maximum relative movement that the positioning element can withstand during detection.

[0030] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] The core of this application is to provide a fully automated production method for an embodied intelligent robot's adaptive assembly, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes: S101. During the process of transporting the pallet by the four-way shuttle, obtain the limiting direction of each positioning component of the pallet, the contact state with the main body to be assembled, the impact direction of the four-way shuttle, the calibrated contact posture, the lateral hole position of the main body to be assembled on the pallet, and the initial assembly trajectory.

[0032] In this step, since there are various types of positioning components, the limiting direction has different meanings for different types of positioning components. For example, for a fixed support block, the limiting direction is the normal direction from its support surface to the bottom surface of the main body to be assembled; for a lateral positioning component, the limiting direction is the normal direction from the lateral positioning surface to the lateral reference surface of the main body to be assembled; for an end positioning component, the limiting direction is the normal direction from the end positioning surface to the end face of the main body to be assembled. It should be noted that the limiting direction is not the direction of the robot's movement command, but rather a geometric reference used to determine the impact component, contact constraint, and trajectory correction direction.

[0033] In this application, a pallet coordinate system is established with the center of the locking hole on the pallet as the origin. The length direction, width direction and the normal of the main positioning surface of the pallet are respectively used as the X-axis, Y-axis and Z-axis. The limiting direction, track changing impact direction, hole position and assembly trajectory can all be uniformly converted to this pallet coordinate system for expression, thereby avoiding the problem of incorrect direction judgment caused by mixing the shuttle coordinate system, robot base coordinate system and the coordinate system of the body to be assembled.

[0034] In S101, the executing entity can be a robot controller. To obtain the contact state, each positioning component can be equipped with a corresponding thin-film pressure sensor, piezoelectric sensor, or microswitch. When the pressure value is not less than the contact threshold and remains for a first duration, the corresponding positioning component is determined to be in a contact state; when the pressure value is not greater than the release threshold and remains for a second duration, the corresponding positioning component is determined to be in a disengaged state, wherein the release threshold is less than the contact threshold. Thus, the contact state obtained in S101 can filter out transient triggering caused by track-changing vibration.

[0035] To obtain the track-changing impact direction, the four-way shuttle collects acceleration during the track-changing process using an onboard triaxial accelerometer. The impact acquisition period can be determined by combining the pallet attitude and lifting drive current. The triaxial acceleration during this period is converted to the pallet coordinate system, and after removing the normal component of the main positioning surface, the remaining acceleration is accumulated over time. The resulting velocity increment direction within the pallet plane is taken as the track-changing impact direction. This direction is used in S102 to screen positioning components that have received effective endoaxial impact.

[0036] The lateral hole position refers to the calibrated hole center, hole axis direction, and hole diameter data of the lateral mounting hole in the coordinate system of the body to be assembled. The initial assembly trajectory is the trajectory of the reversing wheel moving from the part picking position to the lateral mounting hole and completing the insertion when all positioning parts are in the calibrated state. Therefore, the lateral hole position is not the actual hole position directly detected after the track change, but the calibrated hole position that undergoes coordinate transformation according to the assembly reference posture of the body to be assembled.

[0037] S102. Based on the restricted direction, contact state, and track-changing impact direction, determine the main positioning group, the reference positioning group, and the positioning component to be inspected that detaches from the end face after track changing.

[0038] In S102, the preset positioning category to which each positioning component belongs is identified based on the limiting direction. For example, three fixed support blocks whose limiting direction is consistent with the normal of the main positioning surface and which remain in contact during track changing are identified as the main positioning group; two lateral positioning components whose limiting direction is consistent with the normal of the lateral reference surface and which remain in contact are identified as the reference positioning group; and end positioning components whose limiting direction is projected onto the track changing impact direction and reaches the direction threshold, which change from contact to separation after the track changing impact and have not been reconnected by the time they reach the assembly station are identified as positioning components to be inspected.

[0039] like Figure 2As shown, the pallet 100 in the top view is used to support the thin-walled box-shaped main body 200 to be assembled. The main positioning group 120 in the bottom side view includes a first support block 121, a second support block 122, and a third support block 123, the upper surfaces of the three support blocks together forming the main positioning surface; the reference positioning group includes a first lateral positioning member 131 and a second lateral positioning member 132, the positioning surfaces of the two lateral positioning members together forming a lateral reference surface; the positioning member to be inspected 140 is a retractable end stop located at the end of the main body 200 to be assembled. This structural example is used to specifically illustrate the physical correspondence of the three types of positioning structures determined in S102 on the pallet 100.

[0040] exist Figure 2 In the example shown, the normal direction of the main positioning surface is the Z-direction of the pallet coordinate system, the normal direction of the lateral reference surface is the Y-direction, and the intersection direction of the main positioning surface and the lateral reference surface is the X-direction; the positioning component 140 to be inspected restricts the movement of the body 200 to be assembled along the X-direction. When the track-changing impact has a component in the X-direction, the body 200 to be assembled may detach from the positioning component 140 to be inspected, while the main positioning group 120 and the reference positioning group still retain the Z-direction constraint and the Y-direction constraint, respectively, so that the subsequent S104 can perform single-degree-of-freedom detection along the X-direction.

[0041] The main positioning group 120 consists of three fixed support blocks that support the bottom surface of the main body 200 to be assembled. The positioning surfaces of the three fixed support blocks together determine the main positioning surface. The reference positioning group consists of two lateral positioning components spaced apart along the side of the main body 200 to be assembled. The positioning surfaces of the two lateral positioning components together determine the lateral reference surface. The positioning component 140 to be inspected is a collapsible lateral positioning component that, during this rail-changing transportation, has detached from its end face after being impacted and has not yet regained contact upon reaching the assembly station. Its "to be inspected" status only indicates the current inspection state and does not change its mechanical affiliation as an lateral positioning component. The main positioning group 120 and the reference positioning group retain bottom surface constraints and side surface constraints, respectively. The positioning component 140 to be inspected corresponds to the remaining translational degrees of freedom along the intersection of the two positioning surfaces.

[0042] It should be noted that the structural affiliation of the main positioning group 120 and the reference positioning group has been determined during the pallet design. The identification process described above is used to confirm that the corresponding positioning component meets the main positioning or reference positioning conditions during this transportation process, and does not arbitrarily change the mechanical grouping of the positioning components.

[0043] S103. Lock the tray so that the main positioning group and the reference positioning group respectively abut against the main positioning surface and the lateral reference surface of the receiving assembly body.

[0044] In this step, the four-way shuttle delivers the pallet to the assembly station, and the station locking pin is inserted into the locking hole of the pallet. Once the locking pin is in place, the embodied intelligent robot clamps the pre-set clamping area of ​​the main body to be assembled and applies a holding force along the normal direction of the main positioning surface, so that the pressure of the three support blocks is between the contact threshold and the holding upper limit, while keeping the two lateral positioning parts in contact.

[0045] Optionally, when the four-way shuttle is not equipped with a pallet attitude sensor, the transient tilt direction of the pallet can be estimated by the pressure change sequence of the three support blocks during the track switching, and the consistency of the transient tilt direction with the track switching impact direction determined by the vehicle acceleration can be verified. When the two are inconsistent, the pallet is marked as having an abnormal attitude, instead of directly attributing the contact separation to the end impact.

[0046] The problem addressed by the aforementioned identification and locking process is that when the endocentric positioning component detaches, if the main positioning surface or lateral reference surface has also lost its constraint, the position detected by the robot along the endocentric direction will simultaneously include the tilt or deflection of the main body to be assembled. By first confirming that the main positioning group and the reference positioning group are in contact, and then locking the pallet and maintaining the corresponding pressure range, the movable direction of the main body to be assembled can be converged to the single direction corresponding to the positioning component under inspection, so that the subsequent detection position can truly reflect the endocentric reference change.

[0047] S104. During the contact process, the embodied intelligent robot holds the main body to be assembled and sequentially contacts, separates from, and contacts the positioning component to be inspected in the reverse direction along the intersection of the main positioning surface and the lateral reference surface, recording the contact sequence and detection force to form bidirectional detection data.

[0048] In one possible implementation, S104 can be achieved through the following process: Step 41: Determine the direction of the intersection line with positive and negative directions based on the normal of the main positioning surface, the normal of the lateral reference surface, the limiting direction of the positioning component to be inspected, and the impact direction of the track changing.

[0049] Specifically, step 41 can be achieved through the following process: cross-product the normal of the main positioning surface and the normal of the lateral reference surface to obtain the contact tangential direction of the main body to be assembled when the main positioning group and the reference positioning group are in contact; project the limiting direction of the positioning part to be inspected onto the contact tangential direction to obtain the candidate direction; when the projection of the track changing impact direction onto the candidate direction is positive, the candidate direction is determined as positive, and the opposite direction of the candidate direction is determined as negative, forming the intersection direction.

[0050] For example, the normal to the main positioning surface is denoted as n1, and the normal to the lateral reference surface is denoted as n2. The cross product of n1 and n2 is normalized to obtain the contact tangential direction d. Candidate directions are determined based on the projection of the constraint direction of the positioning component under test onto the contact tangential direction d. The positive and negative directions are then determined based on the projection of the track-changing impact direction onto the candidate direction. When the projection of the track-changing impact direction onto the candidate direction is positive, the candidate direction is taken as positive; when the projection is negative, the opposite direction of the candidate direction is taken as positive; when the absolute value of the projection is less than the direction threshold, it indicates that the track-changing impact direction does not match the constraint direction of the positioning component under test, and the robot controller does not initiate this detection. The opposite direction of the positive direction is defined as the reverse direction.

[0051] Step 42: With the main positioning group and the reference positioning group in contact, move the main body to be assembled along the intersection line direction, and determine the contact change moment of the positioning part to be inspected as the segmentation moment.

[0052] The direction of the intersection line is used to limit the detection movement of the main body to be assembled, so that when the main positioning group and the reference positioning group are in contact, the main body to be assembled only moves along the degree of freedom corresponding to the positioning part to be inspected.

[0053] Step 43: According to the segmented time, the movement process of the main body to be assembled is divided into a reverse reconnection segment, a forward disengagement segment, and a reverse verification segment.

[0054] Specifically, step 43 is implemented through the following process: The smaller of the retraction stroke of the positioning component to be inspected and the interference-free stroke of the embodied intelligent robot is determined as the detection stroke; the upper limit of the force on the end face of the main body to be assembled is determined as the upper limit of the detection force; at the beginning of each detection segment, the end displacement is re-accumulated, and the detection force along the intersection direction and the contact state of the reference positioning group are monitored; under the conditions that the end displacement is less than the detection stroke, the detection force is less than the upper limit of the detection force, and the contact state is that the reference positioning group remains in contact, the main body to be assembled is moved in the opposite direction of the intersection direction until it contacts the positioning component to be inspected, forming a reverse reconnection segment; from the end of the reverse reconnection segment, the main body to be assembled is moved in the positive direction of the intersection direction until it disengages from the positioning component to be inspected, forming a positive disengagement segment; from the end of the positive disengagement segment, the main body to be assembled is again moved in the opposite direction of the intersection direction until it contacts the positioning component to be inspected, forming a reverse verification segment. The upper limit of the detection force is not greater than the allowable force on the end face of the main body to be assembled.

[0055] Step 44: Arrange the contact sequence, detection force, end displacement, and end pose of the embodied intelligent robot corresponding to the reverse reconnection segment, forward disengagement segment, and reverse verification segment according to the sampling time to form bidirectional detection data.

[0056] In this step, the contact state of the positioning component under test, the detection force in the intersection direction, and the end-effector displacement and pose of the embodied intelligent robot, collected in the reverse reconnection section, the forward disengagement section, and the reverse verification section, can be arranged according to the sampling time to form bidirectional detection data. During the data formation process, the pressure value of the main positioning group and the contact state of the reference positioning group can also be correlated to improve the richness of the data categories.

[0057] like Figure 3 As shown, the assembly body 200, while maintaining contact with the main positioning group and the reference positioning group, sequentially performs reverse reconnection, forward disengagement, and reverse verification along the intersection line direction. In the reverse reconnection phase, the embodied intelligent robot moves the assembly body 200 in the reverse direction until the inspection positioning component 140 changes from a disengaged state to a contact state. In the forward disengagement phase, the embodied intelligent robot moves the assembly body 200 in the forward direction until the inspection positioning component 140 changes from a contact state to a disengaged state. In the reverse verification phase, the embodied intelligent robot again moves the assembly body 200 in the reverse direction until the inspection positioning component 140 changes from a disengaged state to a contact state for the second time. The robot controller records the contact state, detection force, end-effector displacement, and end-effector pose of the inspection positioning component 140 at a uniform sampling time within the three detection phases, and correlates this with the pressure value of the main positioning group and the contact state of the reference positioning group to form bidirectional detection data.

[0058] At the start of each detection segment, the robot controller resets the cumulative end displacement value to zero and simultaneously monitors the detection force in the intersection direction, the pressure value of the main positioning group, and the contact status of the reference positioning group.

[0059] For example, in the reverse reconnection section, the robot moves the assembly body at low speed in the opposite direction until the inspection positioning part changes from a disengaged state to a contact state, and records the first contact time t1, the first contact force F1, and the first end pose T1. After the first contact, the robot stops pressing in, thereby avoiding the elastic end block from yielding beyond the calibrated range.

[0060] In the forward disengagement phase, the robot moves the main body to be assembled from the first end pose T1 in the forward direction until the positioning part 140 to be inspected changes from the contact state to the disengagement state, and records the disengagement time t2, the disengagement force Fo and the disengagement end pose To.

[0061] In the reverse verification section, the robot moves the assembly body in the reverse direction from the detached end-effector pose To until the inspection positioning part changes from the detached state to the contact state for the second time. The second contact time t3, the second contact force F2, and the second end-effector pose T2 are recorded.

[0062] The robot controller arranges the contact state, detection force, end effector displacement, and end effector pose within the three detection segments into bidirectional detection data according to a unified sampling clock. When any detection segment reaches the upper limit of detection stroke or detection force before the expected contact change occurs, or when the reference positioning group disengages, the detection is terminated and the bidirectional detection data for this period is marked as invalid; this invalid mark is used to determine the invalid detection state in the subsequent processes described above.

[0063] The bidirectional detection data is specifically represented as a data sequence arranged according to the sampling time. Each sampling item includes at least the robot end-effector pose, the cumulative movement of the current detection segment, the detection force, the contact state of the positioning component under test, and the contact state of the reference positioning group; it may also include the pressure values ​​of each support block in the main positioning group. Thus, the bidirectional detection data records not only the contact position and contact sequence of the positioning component under test, but also whether the main positioning constraints and lateral constraints remain effective when the contact position is formed.

[0064] Optionally, when the difference between two contact positions is close to the reconnection threshold, or the difference between two contact forces is close to the force difference threshold, a forward disengagement and reverse reconnection can be added under the condition that the detection stroke and detection force limit are not exceeded, and the median value of the three contact positions is used as the verified contact position; additional reconnection is only performed when critical data appears, and the number of detections is not increased on each tray.

[0065] The problem addressed by the three-stage detection method is that a single contact can only obtain the current end-face position and cannot determine whether the position is a stable new positioning reference or an accidental position caused by elastic deformation, friction jamming, or robot clamping deformation. By sequentially establishing contact, releasing contact, and reconstructing contact, position repeatability and force loops can jointly eliminate accidental contact, thus providing a verifiable data basis for distinguishing between body displacement and positioning slippage.

[0066] S105. Based on the bidirectional detection data, the calibrated contact pose, and the end pose of the embodied intelligent robot during reverse contact, forward separation, and re-reverse contact, determine the positioning state and assembly reference pose of the positioning component to be inspected.

[0067] In one possible implementation, S105 can be achieved through the following process: Step 501: Extract the detection force at the first contact, the detection force at the separation, and the detection force at the second contact from the bidirectional detection data, and associate them with the corresponding end poses to form reconnection data.

[0068] Step 502: Based on the reconnection data, determine the reconnection difference between the two contact positions, the force difference between the two contact forces, and the loop difference enclosed by the contact and disengagement processes.

[0069] Step 503: Determine the position difference between the end pose at the second contact and the calibrated contact pose as the reference difference.

[0070] Step 504: Combine the reconnection difference, force difference, lap difference, and reference difference to form a positioning feature.

[0071] Step 505: Determine the positioning state of the positioning component to be inspected based on the positioning features, and select the assembly reference pose from the calibrated contact pose and the end pose at the second contact based on the positioning state.

[0072] Specifically, step 505 is explained in the following three cases: When the reconnection difference is not greater than the reconnection threshold, the force difference is not greater than the force difference threshold, the return line difference is not greater than the return line threshold, and the reference difference is not greater than the preset reference threshold, the positioning state is determined as the main body is off-position, and the calibrated contact pose is determined as the assembly reference pose; when the reconnection difference is not greater than the reconnection threshold, the force difference is not greater than the force difference threshold, the return line difference is not greater than the return line threshold, and the reference difference is greater than the preset reference threshold, the positioning state is determined as positioning slip, and the end pose at the second contact is determined as the assembly reference pose; when the reconnection difference is greater than the reconnection threshold, the force difference is greater than the force difference threshold, or the return line difference is greater than the return line threshold, the positioning state is determined as detection invalid, and the determination of the assembly reference pose is stopped.

[0073] In the above process, the reconnection difference represents the distance difference between the two reverse contact positions along the intersection direction, the force difference represents the difference in the detection force during the two reverse contacts, the return line difference represents the area enclosed by the force-displacement curves of the reverse reconnection process and the forward separation process within the common displacement interval, and the reference difference represents the position difference between the reconnection pose and the calibration pose along the limiting direction of the positioning component 140 under test; the above four quantities together constitute the positioning feature.

[0074] The above process extracts the data corresponding to the first contact, disengagement, and second contact from the bidirectional detection data to form reconnection data. The absolute value of the position difference between the first end pose T1 and the second end pose T2 on the contact tangential direction d is determined as the reconnection difference Δp, and the absolute value of the difference between the first contact force F1 and the second contact force F2 is determined as the force difference ΔF.

[0075] The area enclosed by the detection force-displacement curves of the reverse reconnection section and the forward disconnection section within the common displacement interval is defined as the hysteresis difference Ah. Specifically, the common displacement interval can be divided into multiple sampling intervals, and the product of the detection force difference between the two curves in each sampling interval and the interval length is calculated, and the products are summed. The hysteresis difference Ah is used to characterize whether there is jamming, elastic hysteresis, or reference positioning instability between the two contacts.

[0076] The absolute value of the positional difference between the second end pose T2 and the calibrated contact pose T0 in the restricted direction of the positioning component 140 under test is determined as the reference difference Δb. The reconnection difference Δp, force difference ΔF, lap difference Ah, and reference difference Δb together constitute the positioning characteristics. Each threshold can be determined by multiple calibration detection data when the positioning component is not slipping, for example, by using the sum of the calibration mean and a preset multiple of the standard deviation as the corresponding threshold.

[0077] When the reconnection difference Δp is not greater than the reconnection threshold, the force difference ΔF is not greater than the force difference threshold, the return line difference Ah is not greater than the return line threshold, and the reference difference Δb is not greater than the reference threshold, it indicates that the two contacts can repeatedly reach the original calibration reference. The controller determines the positioning state as the main body is out of position and determines the calibration contact posture T0 as the assembly reference posture.

[0078] Among them, "body displacement" means that the body to be assembled leaves the end-to-calibrated position after the track-changing impact, but the positioning part to be inspected is still in the calibrated position; "positioning slippage" means that the positioning part to be inspected can be repeatedly contacted, but its stable contact position is offset from the calibrated contact position; "invalid detection" means that the position, force, or contact loop of the two contacts is not repeatable. The assembly reference pose is the pose of the body to be assembled used to transform the lateral hole position to the pallet coordinate system: when the body is displaced, the calibrated contact pose T0 is taken; when positioning slippage occurs, the second end pose T2 is taken; when the detection is invalid, it is not determined.

[0079] When the reconnection difference Δp, force difference ΔF, and return line difference Ah are not greater than their respective thresholds, while the reference difference Δb is greater than the reference threshold, it indicates that the two contacts can repeatedly reach the same new position, but the position deviates from the original calibration reference. The controller determines the positioning state as positioning slip and determines the second end pose T2 as the assembly reference pose.

[0080] When any of the reconnection difference Δp, force difference ΔF, or homing difference Ah exceeds the corresponding threshold, or when the bidirectional detection data contains an invalid mark, the controller will determine the positioning status as invalid detection, will not output the assembly reference pose, and will transfer the pallet to the manual verification station or the repositioning station.

[0081] For example, the reconnection threshold can be 0.10 mm, the force difference threshold can be 3 N, the lap line threshold can be 0.25 N·mm, and the reference threshold can be 0.30 mm. If the difference between two contact positions is 0.04 mm, the difference in contact force is 1.2 N, the lap line difference is 0.08 N·mm, and the second contact position is offset from the calibration position by 0.65 mm, then the positioning state is determined to be positioning slippage. The above values ​​are only exemplary values ​​used to illustrate the determination process and do not constitute a limitation on the threshold range.

[0082] It should be noted that each threshold is determined based on the calibration data of the main body to be assembled, the positioning components, and the sensors. During calibration, a box-shaped main body to be assembled, made of the same material and wall thickness as the actual production batch, is used. The main positioning group, reference positioning group, and positioning components to be inspected are used, all with the same structure, material, and rigidity as the production pallet. A torque sensor actually used is installed at the end effector of the embodied intelligent robot, and multiple impact-free repeated detections are performed using the same sampling frequency, detection speed, detection force control method, and clamping posture as in the production process. Based on the difference between the two contact positions, the difference in contact force, the hysteresis loop difference, and the position difference between the re-contact pose and the calibrated contact pose obtained from each detection, the mean and dispersion of the corresponding data are calculated. Based on the allowable assembly position tolerance, the allowable clearance of the positioning components, and the sensor measurement error, the reconnection threshold, force difference threshold, hysteresis loop threshold, and reference threshold are determined. These thresholds are written into the robot controller before production and recalibrated after changing the box material, positioning components, or sensors.

[0083] Optionally, when the main body to be assembled is a thin-walled part that enters the assembly line directly after heat treatment, the theoretical expansion component caused by temperature change in the reference difference can be deducted according to the pallet temperature and the linear expansion coefficient of the material of the main body to be assembled, and then the remaining component is compared with the reference threshold. This temperature correction only changes the comparison value of the reference difference and does not change the formation method of the reconnection difference, force difference and return line difference.

[0084] In this application, the return of the main body to the end contact position does not necessarily mean that the end positioning component is normal. This application can first confirm whether the new contact position is stable by reconnection difference, force difference, and return line difference, and then determine whether the stable position is still at the original calibration reference by the reference difference. This can separate the case where the main body is pushed away but the positioning component is normal from the case where the positioning component has slipped. For contacts that cannot be repeated, the assembly reference pose is stopped from being output to avoid continuing to perform automatic insertion using unreliable positions.

[0085] S106. Based on the positioning state, assembly reference pose, and lateral hole position, determine the assembly trajectory of the reversing wheel, which is the part to be assembled, and insert the reversing wheel into the lateral mounting hole of the main body to be assembled along the assembly trajectory.

[0086] In one possible implementation, the process of determining the assembly trajectory of the reversing wheel, which is the part to be assembled, in S106 includes the following steps: Step 61: Calculate the positional difference between the assembly reference pose and the calibration contact pose in the limiting direction of the positioning part to be inspected, and obtain the candidate correction amount.

[0087] Step 62: When the positioning state is that the main body is out of position, zero is determined as the hole position correction amount; when the positioning state is that the positioning is slippage, the candidate correction amount is determined as the hole position correction amount.

[0088] Among them, the candidate correction amount represents the signed position difference between the assembly reference pose and the calibration contact pose in the restricted direction of the positioning part to be inspected; the hole position correction amount represents the actual trajectory translation amount selected from zero and the candidate correction amount according to the positioning state.

[0089] Step 63: Based on the position of the reference positioning group and the lateral hole position, determine the correction start point and correction end point on the initial assembly trajectory, and determine the trajectory segment between the correction start point and the correction end point as the correction range.

[0090] The lateral hole positions include at least the hole center, hole axis direction, and hole diameter of the lateral mounting hole in the coordinate system of the body to be assembled. The lateral hole positions are transformed to the pallet coordinate system according to the assembly reference pose to obtain the target hole positions used in this assembly.

[0091] Specifically, step 63 can be achieved through the following process: construct an inlet surface perpendicular to the main positioning surface by passing through the material-side edge points of the two positioning components in the reference positioning group; construct an orifice surface perpendicular to the axis of the lateral mounting hole by passing through the center of the orifice of the lateral mounting hole; determine the intersection of the initial assembly trajectory with the inlet surface and the orifice surface as the correction start point and correction end point, respectively, and determine the initial assembly trajectory between the correction start point and the correction end point as the correction range.

[0092] Step 64: Based on the position of each trajectory point within the correction range, translate and correct each trajectory point along the limiting direction of the positioning component to be inspected, and retain the initial assembly trajectory outside the correction range to obtain the assembly trajectory of the reversing wheel.

[0093] like Figure 4 As shown, the above process first calculates the signed position difference between the assembly reference pose and the calibrated contact pose in the restricted direction of the positioning part to be inspected, and obtains the candidate correction amount. When the positioning state is that the main body is out of position, the hole position correction amount δ is set to zero; when the positioning state is that the positioning is sliding, the candidate correction amount is determined as the hole position correction amount δ.

[0094] By referencing the incoming edge points of the two lateral positioning components in the positioning group, an inlet surface P1 perpendicular to the main positioning surface is constructed; by passing through the center of the orifice of the lateral mounting hole 210, an orifice surface P2 perpendicular to the axis of the lateral mounting hole is constructed. Following the feed sequence of the reversing wheel, the trajectory point where the initial assembly trajectory first crosses the inlet surface P1 is determined as the correction starting point, and the trajectory point where the initial assembly trajectory first reaches the orifice surface P2 is determined as the correction ending point.

[0095] The correction range is the initial assembly trajectory segment between the correction start point and the correction end point, not the solid area on the pallet; the inlet surface P1 is used to characterize the position where the reversing wheel begins to enter the channel defined by the reference positioning group, and the orifice surface P2 is used to characterize the position where the reversing wheel positioning shaft reaches the orifice of the lateral mounting hole 210. By expressing the correction range with the intersection of two geometric surfaces and the trajectory, the controller can directly determine the continuous segment from the existing trajectory point sequence that needs to have the hole position correction amount added.

[0096] The initial assembly trajectory between the correction start point and the correction end point is defined as the correction range. To avoid positional jumps at the correction start point, a transition distance Lt is intercepted along the initial assembly trajectory from the correction start point to obtain the transition end point. The trajectory points between the correction start point and the transition end point are adjusted according to the ratio of their cumulative arc length to the transition distance Lt, so that the translation amount increases from zero to the hole position correction amount δ. The trajectory points between the transition end point and the correction end point maintain the translation hole position correction amount δ.

[0097] The above translation is performed on each trajectory point within the correction range along the restricted direction of the positioning part 140 to be inspected, while keeping the trajectory points outside the correction range unchanged, thus obtaining the assembly trajectory of the reversing wheel. The velocity continuity processing is performed on the translated trajectory points so that the velocity and acceleration at the transition endpoint satisfy the robot motion constraints.

[0098] For example, in the positioning and sliding state, the hole position correction amount δ is 0.65 mm, the trajectory length from the correction start point to the correction end point is 80 mm, and the transition distance Lt is 20 mm. The robot increases the translation amount from 0 to 0.65 mm in the first 20 mm, and maintains a translation amount of 0.65 mm in the next 60 mm; thus, after the reversing wheel passes through the channel defined by the reference positioning group, its positioning axis corresponds to the lateral mounting hole 210 that is offset with the body 200 to be assembled.

[0099] It should be noted that, in the positioning and sliding state, the controller determines the hole position correction amount based on the signed position difference between the assembly reference pose and the calibrated contact pose in the restricted direction of the positioning part to be inspected. The trajectory length, transition distance, and hole position correction amount between the correction start point and the correction end point are not predetermined as fixed values, but are determined based on the passable clearance between the reference positioning groups, the allowable deviation between the reversing wheel positioning shaft and the lateral mounting hole, the actual sliding amount of the positioning part to be inspected, and the position repeatability accuracy of the embodied intelligent robot. The transition distance is calculated by the robot trajectory controller based on the maximum allowable rate of position change and the maximum allowable acceleration, so that the translation amount at the correction start point continuously increases from zero to the hole position correction amount, and the hole position correction amount is maintained after the transition section ends until the reversing wheel reaches the lateral mounting hole.

[0100] Optionally, when two lateral mounting holes are provided on the main body to be assembled, which are spaced apart along the intersection direction, and the two holes have corresponding assembly reference poses, the small deflection amount of the main body to be assembled around the normal of the main positioning surface can be determined according to the difference between the two hole position corrections, and a rotation correction that varies with the feed distance is added only to the trajectory segment between the two holes; when only a single reversing wheel is assembled or the difference between the two hole position corrections does not exceed the rotation threshold, the rotation correction is not performed.

[0101] The problem addressed by local trajectory correction is that directly adding the hole position correction amount to the entire initial assembly trajectory will change the robot's safe trajectory before leaving the part pick-up position and entering the tooling channel; adding correction only at the hole opening will cause a position jump. By limiting the correction range by the entrance surface and the hole opening surface, and gradually increasing the translation amount after the correction starting point, the existing obstacle avoidance relationship away from the tooling area can be maintained, while allowing the reversing wheel to continuously approach the offset lateral mounting hole after entering the reference positioning group.

[0102] In S106, inserting the reversing wheel into the lateral mounting hole of the body to be assembled along the assembly trajectory can be achieved through the following process: Step 65: When the main positioning group and the reference positioning group are in contact and the positioning state is not invalid, control the station clamping component to clamp the body to be assembled and keep the body to be assembled in the assembly reference pose; after detecting the clamping position signal of the station clamping component, control the embodied intelligent robot to release the body to be assembled and clamp the reversing wheel; Step 66: Move the reversing wheel along the assembly trajectory to the reference positioning group so that the positioning axis of the reversing wheel corresponds to the lateral mounting hole, forming an entry pose; Step 67: Move the reversing wheel from the entry pose along the normal of the lateral reference plane and limit the force of the reversing wheel along the limiting direction of the positioning component to be inspected; Step 68: When the positioning axis of the reversing wheel enters the lateral mounting hole and the insertion depth reaches the predetermined depth, stop moving the reversing wheel.

[0103] The entry pose is the robot end-effector pose when the reversing wheel reaches the reference positioning group along the corrected assembly trajectory, and the reversing wheel positioning axis corresponds to the lateral mounting hole in the hole axis direction. The entry pose is neither the part-picking pose nor the final insertion pose, but rather a transition pose where movement along the trajectory is switched to restricted feed along the hole axis. The hole axis direction of the lateral mounting hole 210 is parallel to the normal direction of the lateral reference plane.

[0104] Furthermore, the workstation clamping component is set at the assembly station and is independent of the embodied intelligent robot. The controller of the workstation clamping component is communicatively connected to the controller of the embodied intelligent robot. After the embodied intelligent robot completes bidirectional detection of the main body to be assembled, the workstation clamping component holds the main body to be assembled in the assembly reference posture and outputs a clamping signal. In response to the clamping signal, the embodied intelligent robot releases the main body to be assembled and switches to clamping the reversing wheel.

[0105] After the assembly trajectory is determined, the station clamping component applies clamping force from above the main body to be assembled or from the side opposite to the positioning component to be inspected. During the clamping period, the embodied intelligent robot continues to hold the main body to be assembled to avoid changing the assembly reference posture during the clamping process. When the positioning sensor of the station clamping component outputs a clamping positioning signal, the main positioning group, the reference positioning group, and the station clamping component jointly maintain the assembly reference posture of the main body to be assembled. The embodied intelligent robot then releases the main body to be assembled and switches from the main body gripper to the reversing wheel gripper through an automatic gripper-changing mechanism to clamp the reversing wheel for subsequent insertion. The station clamping component is located at the assembly station and is independent of the embodied intelligent robot, but their controllers are communicatively connected.

[0106] In this embodiment, starting from the entry pose, the embodied intelligent robot moves the reversing wheel along the normal direction of the lateral reference plane, and limits the insertion force in the hole axis direction and the lateral force in the limiting direction of the positioning component to be inspected. Feeding stops when the positioning shaft enters the lateral mounting hole and the insertion depth reaches the target depth. If any force exceeds the corresponding upper limit, the robot retreats a preset distance in the opposite direction of the hole axis and stops the assembly.

[0107] In one example, the main body to be assembled is a thin-walled box-shaped reversing wheel mounting housing. The reversing wheel includes a wheel body, a bracket, and a positioning shaft extending from the side of the bracket. Lateral mounting holes are formed in the side wall of the housing or in bearing seats on the side wall. After the positioning shaft enters the lateral mounting hole, the robot continues to move the bracket to the assembly surface, where it is then fastened by a subsequent station.

[0108] Optionally, after the positioning axis reaches the target insertion depth, the robot can perform a return release in the opposite direction of the hole axis that is less than the disengagement stroke, and collect the axial force and the lateral residual force along the limiting direction of the positioning part under inspection during the return process; when the lateral residual force exceeds the residual force threshold, the reversing wheel is disengaged and the assembly reference pose is rechecked; when the lateral residual force does not exceed the residual force threshold, the robot returns to the target insertion depth and completes the fastening handover.

[0109] In this application, if the robot releases the body to be assembled directly after completing the detection, the main positioning group and the reference positioning group can only provide contact constraints and cannot guarantee that the body to be assembled will maintain its endocentric position during the robot's gripper change. The station clamping component first locks the assembly reference pose, and then allows the robot to release the body and clamp the reversing wheel, so that the assembly reference obtained by detection can be continued into the insertion process; during insertion, both axial and lateral forces are restricted, which can prevent the positioning axis from continuously compressing the thin-walled hole wall even though it has reached the insertion depth.

[0110] This application, through steps S101 to S106, after the main body to be assembled is separated from the end positioning component due to the track-changing impact, firstly, the main positioning group and the reference positioning group, which are still in contact, are used to retain the main positioning surface and the lateral reference surface. Then, the positioning component to be inspected is repeatedly probed along the intersection of the two positioning surfaces, thereby distinguishing the temporary displacement of the main body to be assembled from the relative sliding of the positioning component to be inspected relative to the pallet. When determining the positioning slippage, the lateral hole position and the local trajectory of the reversing wheel before entering the lateral mounting hole are corrected by re-contacting the corresponding assembly reference position, thereby avoiding the use of the distorted end calibration reference and completing the adaptive insertion after track-changing transportation without adding a rigid fully constrained fixture.

[0111] Figure 5 A schematic diagram of a fully automated production embodied intelligent robot adaptive assembly system provided in this application embodiment, referring to... Figure 5 The system may include: The acquisition module 51 is used to acquire the limiting direction of each positioning component of the pallet, the contact state with the main body to be assembled, the track changing impact direction of the four-way shuttle, the calibrated contact posture, the lateral hole position of the main body to be assembled on the pallet and the initial assembly trajectory during the process of the four-way shuttle changing the track transport pallet.

[0112] The first determining module 52 is used to determine the main positioning group, the reference positioning group, and the positioning component to be inspected that detaches from the end face after the track change, based on the limiting direction, the contact state, and the track-changing impact direction.

[0113] The locking module 53 is used to lock the tray so that the main positioning group and the reference positioning group respectively abut against the main positioning surface and the lateral reference surface of the receiving assembly body.

[0114] The clamping module 54 is used to clamp the body to be assembled by the embodied intelligent robot during the contact process, and sequentially contact the positioning part to be inspected in the reverse direction, separate it in the forward direction, and contact it again in the reverse direction along the intersection of the main positioning surface and the lateral reference surface, and record the contact sequence and detection force to form bidirectional detection data.

[0115] The second determining module 55 is used to determine the positioning state and assembly reference posture of the positioning component to be inspected based on the bidirectional detection data, the calibrated contact posture, and the end posture of the embodied intelligent robot when it makes reverse contact, forward separation, and reverse contact again.

[0116] The insertion module 56 is used to determine the assembly trajectory of the reversing wheel, which is the part to be assembled, according to the positioning state, assembly reference pose and lateral hole position, and insert the reversing wheel into the lateral mounting hole of the main body to be assembled along the assembly trajectory.

[0117] The fully automated production embodied intelligent robot adaptive assembly system of this application embodiment is used to implement the aforementioned fully automated production embodied intelligent robot adaptive assembly method. Therefore, the specific implementation of the fully automated production embodied intelligent robot adaptive assembly system can be found in the embodiment section of the fully automated production embodied intelligent robot adaptive assembly method above. The specific implementation can be referred to the description of the corresponding embodiment, and will not be repeated here.

[0118] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the fully automated production adaptive assembly method of the embodied intelligent robot described above.

[0119] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the fully automated production adaptive assembly method for embodied intelligent robots described above.

[0120] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0121] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the fully automated production embodied intelligent robot adaptive assembly method described above.

[0122] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0123] The above provides a detailed description of the fully automated production method and system for adaptive assembly using an embodied intelligent robot. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A fully automated production method for adaptive assembly using an embodied intelligent robot, characterized in that, include: During the process of transporting the pallet by the four-way shuttle, the limiting direction of each positioning component of the pallet, the contact state with the main body to be assembled, the impact direction of the four-way shuttle, the calibrated contact posture, the lateral hole position of the main body to be assembled on the pallet, and the initial assembly trajectory are obtained. Based on the restricted direction, contact state, and track-changing impact direction, determine the main positioning group, the reference positioning group, and the positioning component to be inspected that detaches from the end face after track changing; The tray is locked so that the main positioning group and the reference positioning group respectively abut against the main positioning surface and the lateral reference surface of the receiving assembly body; During the contact process, the embodied intelligent robot holds the main body to be assembled and sequentially contacts, separates from, and contacts the positioning part to be inspected in the opposite direction along the intersection of the main positioning surface and the lateral reference surface, recording the contact sequence and detection force to form bidirectional detection data. Based on the bidirectional detection data, the calibrated contact pose, and the end pose of the embodied intelligent robot during reverse contact, forward disengagement, and re-reverse contact, the positioning state and assembly reference pose of the positioning component to be inspected are determined. Based on the positioning state, assembly reference pose, and lateral hole position, the assembly trajectory of the reversing wheel, which is the part to be assembled, is determined, and the reversing wheel is inserted into the lateral mounting hole of the main body to be assembled along the assembly trajectory.

2. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 1, characterized in that, The assembly body is held by an embodied intelligent robot, and along the intersection of the main positioning surface and the lateral reference surface, it sequentially contacts the positioning component under inspection in the reverse direction, separates from it in the forward direction, and then contacts it again in the reverse direction. The contact sequence and detection force are recorded to form bidirectional detection data, including: Based on the normal of the main positioning surface, the normal of the lateral reference surface, the limiting direction of the positioning component to be inspected, and the impact direction of the track changing, determine the direction of the intersection line with positive and negative directions; With both the main positioning group and the reference positioning group in contact, the main body to be assembled is moved along the intersection direction, and the moment of contact change of the positioning part to be inspected is determined as the segmentation moment. According to the segmented time, the movement process of the main body to be assembled is divided into a reverse reconnection segment, a forward disengagement segment, and a reverse verification segment. The contact sequence, detection force, end displacement, and end pose of the embodied intelligent robot corresponding to the reverse reconnection segment, forward disengagement segment, and reverse verification segment are arranged according to the sampling time to form bidirectional detection data.

3. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 2, characterized in that, The determination of the intersection direction with positive and negative directions based on the normal of the main positioning surface, the normal of the lateral reference surface, the limiting direction of the positioning component to be inspected, and the impact direction of the track changing includes: By cross-product of the normal of the main positioning surface and the normal of the lateral reference surface, the contact tangent of the body to be assembled is obtained when the main positioning group and the reference positioning group are in contact. Projecting the limiting direction of the positioning component under test onto the contact tangent, we obtain the candidate direction; When the projection of the track-changing impact direction onto the candidate direction is positive, the candidate direction is determined as positive, and the opposite direction of the candidate direction is determined as negative, thus forming the intersection direction.

4. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 2, characterized in that, The process of moving the main body to be assembled is divided into a reverse reconnection segment, a forward disengagement segment, and a reverse verification segment according to the segmented time, including: The smaller value between the retraction stroke of the positioning component to be inspected and the non-interference stroke of the embodied intelligent robot is determined as the detection stroke, and the upper limit of the force on the end face of the main body to be assembled is determined as the upper limit of the detection force. At the beginning of each detection segment, the end displacement is re-accumulated, and the detection force along the intersection direction and the contact status of the reference positioning group are monitored. Under the conditions that the end displacement is less than the detection stroke, the detection force is less than the upper limit of the detection force, and the contact state is that the reference positioning group is in contact, the body to be assembled is moved in the opposite direction of the intersection line until the positioning part to be inspected contacts, forming a reverse reconnection section. Move the main body to be assembled from the end of the reverse reconnection section along the positive direction of the intersection line until the positioning part to be inspected is disengaged, forming the positive disengagement section; From the end of the forward separation section, move the assembly body again in the opposite direction of the intersection line until it contacts the positioning part to be inspected, forming the reverse verification section.

5. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 4, characterized in that, The step of determining the positioning state and assembly reference pose of the component to be inspected based on the bidirectional detection data, the calibrated contact pose, and the end-effector pose of the embodied intelligent robot during reverse contact, forward disengagement, and re-reverse contact includes: The detection force at the first contact, the detection force at the separation, and the detection force at the second contact are extracted from the bidirectional detection data, and associated with the corresponding end poses to form reconnection data. Based on the reconnection data, determine the reconnection difference between the two contact positions, the force difference between the two contact forces, and the loop difference enclosed by the contact and disengagement processes; The positional difference between the end pose at the second contact and the calibrated contact pose is determined as the reference difference; The aforementioned connection difference, force difference, hysteresis difference, and reference difference constitute the positioning features; The positioning state of the positioning component to be inspected is determined based on the positioning features, and the assembly reference pose is selected from the calibrated contact pose and the end pose at the second contact based on the positioning state.

6. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 5, characterized in that, The step of determining the positioning state of the component to be inspected based on the positioning features, and selecting the assembly reference pose from the calibrated contact pose and the end pose at the second contact based on the positioning state, includes: When the reconnection difference is not greater than the reconnection threshold, the force difference is not greater than the force difference threshold, the return line difference is not greater than the return line threshold, and the reference difference is not greater than the preset reference threshold, the positioning state is determined as the main body is off-position, and the calibrated contact posture is determined as the assembly reference posture. When the reconnection difference is not greater than the reconnection threshold, the force difference is not greater than the force difference threshold, the return line difference is not greater than the return line threshold, and the reference difference is greater than the preset reference threshold, the positioning state is determined as positioning slip, and the end pose at the second contact is determined as the assembly reference pose. When the repeat difference is greater than the repeat threshold, the force difference is greater than the force difference threshold, or the lap difference is greater than the lap threshold, the positioning status is determined as invalid detection, and the determination of the assembly reference pose is stopped.

7. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 1, characterized in that, The step of determining the assembly trajectory of the reversing wheel, which is the part to be assembled, based on the positioning state, assembly reference pose, and lateral hole position includes: Calculate the positional difference between the assembly reference pose and the calibration contact pose in the limiting direction of the positioning part to be inspected, and obtain the candidate correction amount; When the positioning state is that the main body is out of position, zero is determined as the hole position correction amount; when the positioning state is that the positioning is slipping, the candidate correction amount is determined as the hole position correction amount. Based on the position of the reference positioning group and the lateral hole position, the correction start point and correction end point are determined on the initial assembly trajectory, and the trajectory segment between the correction start point and the correction end point is determined as the correction range; Based on the position of each trajectory point within the correction range, each trajectory point is translated and corrected along the limiting direction of the positioning component to be inspected, and the initial assembly trajectory outside the correction range is retained to obtain the assembly trajectory of the reversing wheel.

8. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 7, characterized in that, The step of determining the correction start point and correction end point on the initial assembly trajectory based on the position of the reference positioning group and the lateral hole position, and defining the trajectory segment between the correction start point and the correction end point as the correction range, includes: By referring to the material-side edge points of the two positioning components in the positioning group, an inlet surface perpendicular to the main positioning surface is constructed; Create an opening surface perpendicular to the axis of the lateral mounting hole by passing through the center of the opening; The intersections of the initial assembly trajectory with the inlet surface and the orifice surface are respectively determined as the correction start point and correction end point, and the initial assembly trajectory between the correction start point and the correction end point is determined as the correction range.

9. The fully automated production method for adaptive assembly using a unibody intelligent robot according to claim 1, characterized in that, The step of inserting the reversing wheel into the lateral mounting hole of the body to be assembled along the assembly trajectory includes: When the main positioning group and the reference positioning group are in contact and the positioning state is not invalid, the control station clamping component clamps the body to be assembled and keeps the body to be assembled in the assembly reference position. After detecting the clamping signal of the clamping component at the workstation, the intelligent robot is controlled to release the main body to be assembled and clamp the reversing wheel; Move the reversing wheel along the assembly trajectory to the reference positioning group so that the positioning axis of the reversing wheel corresponds to the lateral mounting hole, forming the entrance pose; The reversing wheel is moved along the normal of the lateral reference plane from the entrance position, and the force exerted by the reversing wheel along the limiting direction of the positioning component under test is restricted. When the positioning shaft of the reversing wheel enters the lateral mounting hole and the insertion depth reaches the predetermined depth, the reversing wheel stops moving.

10. A fully automated production system using an embodied intelligent robot for adaptive assembly, characterized in that, include: The acquisition module is used to acquire the limiting direction of each positioning component of the pallet, the contact state with the main body to be assembled, the impact direction of the four-way shuttle changing track, the calibrated contact posture, the lateral hole position of the main body to be assembled on the pallet and the initial assembly trajectory during the process of the four-way shuttle changing track transport pallet. The first determining module is used to determine the main positioning group, the reference positioning group, and the positioning component to be inspected that detaches from the end face after the track change, based on the limiting direction, the contact state, and the track-changing impact direction. A locking module is used to lock the pallet so that the main positioning group and the reference positioning group respectively abut against the main positioning surface and the lateral reference surface of the receiving assembly body; The clamping module is used to clamp the body to be assembled by an embodied intelligent robot during the contact process, and sequentially contact the positioning part to be inspected in the reverse direction, separate it in the forward direction, and contact it again in the reverse direction along the intersection of the main positioning surface and the lateral reference surface, and record the contact sequence and detection force to form bidirectional detection data. The second determining module is used to determine the positioning state and assembly reference posture of the positioning component to be inspected based on the bidirectional detection data, the calibrated contact posture, and the end posture of the embodied intelligent robot when it makes reverse contact, forward separation, and reverse contact again. The insertion module is used to determine the assembly trajectory of the reversing wheel, which is the part to be assembled, based on the positioning state, assembly reference pose and lateral hole position, and insert the reversing wheel into the lateral mounting hole of the main body to be assembled along the assembly trajectory.

Images