Collaborative robot intelligent loading and unloading system

Abstract

The application belongs to the technical field of robots, and discloses a kind of collaborative robot intelligent feeding and discharging systems, including station structure perception module, the station structure information of station where collaborative robot executes feeding and discharging task is collected, and station structure information is structured coding, generates station structure constraint matrix;Part task decomposition module, based on station structure constraint matrix, constructs the feeding and discharging task queue of part, and combines preset station beat requirement, converts feeding and discharging task queue into the action instruction set executable by collaborative robot;Path deduction avoidance module, the joint space of collaborative robot is projected to action instruction set, based on station constraint path mapping mechanism, deduction generates candidate path set;Dual-target scheduling of energy consumption and process beat is used, and candidate path is optimized and selected, and the optimal candidate path is selected as the execution path of robot;The safety of collaborative robot feeding and discharging operation is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robots, and more particularly, to a collaborative robot intelligent feeding and discharging system. BACKGROUND

[0002] The existing collaborative robot intelligent feeding and discharging system mainly has the following problems:

[0003] As an important part of intelligent manufacturing and industrial automation, collaborative robots have been widely used in the feeding and discharging links of automobile manufacturing, electronic assembly, metal processing and other scenes. The collaborative robot feeding and discharging system can replace manual repetitive labor to a certain extent and improve production efficiency and operation consistency through the operation of mechanical arms to grab, carry and place parts. However, the existing collaborative robot intelligent feeding and discharging system still has many deficiencies and cannot meet the comprehensive requirements of safety, flexibility and production rhythm in complex work station environments.

[0004] Traditional collaborative robot feeding and discharging methods usually execute tasks in a fixed order, and lack sufficient consideration of the stacking relationship between parts, the occlusion relationship, the spatial conflict and the operation dependency, which may easily lead to operation conflicts or task failures. At the same time, the existing systems lack dynamic control of the action execution order and activation conditions, and cannot flexibly adjust the action timing according to the work station constraints and safety requirements, thereby causing low robot execution efficiency or potential safety hazards.

[0005] In the matching process of the action unit and the work station constraint, the existing methods usually adopt a black box strategy, lack quantitative and interpretation of the action execution compatibility, and cannot guarantee the safety of the action and the coordination of the production rhythm in complex work station environments. Most of the existing technologies only focus on shortening the path execution time or reducing energy consumption, but lack a comprehensive balance between time and energy efficiency, which often leads to high energy consumption at a fast execution speed or low energy consumption that cannot match the production rhythm.

[0006] The existing robot path planning is usually optimized independently and does not fully introduce the beat constraint of the production line, which may easily lead to robot operation time exceeding the cycle of the production line work station, thereby disrupting the connection of complex production processes such as vehicle manufacturing. The existing methods rely on empirical scheduling and lack a unified evaluation index to measure the pros and cons of different candidate paths, which cannot achieve stable and controllable scheduling decisions in complex working conditions.

[0007] In view of the above problems, the present application provides a collaborative robot intelligent feeding and discharging system. SUMMARY

[0008] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purpose, the present application provides the following technical scheme: a collaborative robot intelligent feeding and discharging system, comprising:

[0009] The workstation structure perception module collects the workstation structure information of the workstation where the collaborative robot is performing loading and unloading tasks, and performs structured encoding on the workstation structure information to generate a workstation structure constraint matrix.

[0010] The component task decomposition module constructs a component loading and unloading task queue based on the workstation structure constraint matrix, and transforms the loading and unloading task queue into a set of action instructions that can be executed by the collaborative robot in combination with the preset workstation cycle requirements.

[0011] The path deduction and avoidance module projects the set of motion commands onto the joint space of the collaborative robot, and generates a set of candidate paths based on the workstation constraint path mapping mechanism. It adopts a dual-objective scheduling of energy consumption and process cycle time to optimize and select the candidate path, and selects the best-performing candidate path as the robot's execution path.

[0012] The collaborative safety control module, when collaborative robots and humans are working at the same workstation, infers collaborative conflicts between the collaborative robots and humans based on the workstation structure constraint matrix and action instruction set, and performs safety control on the execution path of the collaborative robots.

[0013] Specifically, the workstation structure information includes the component location information, posture information, and temporary stacking status information of the workstation.

[0014] Specifically, the method for generating the workstation structure constraint matrix includes:

[0015] The collected workstation structure information is standardized. The component position information is transformed into standard three-dimensional coordinates through Min-Max normalization. The attitude information is transformed into a unified quaternion representation using Euler angle quaternion transformation. The temporary stacking state information is transformed into quantifiable constraint parameters.

[0016] A unique identifier is established for each component, and the component's location information, posture information, and temporary stacking status information are associated with this identifier. At the same time, a geometric code is established for the preset workstation space area to form a workstation space element set.

[0017] Based on the geometric relationship between the parts and the workstation space element set, constraint relationships are generated by mapping. The part ID is used as the row index of the matrix, the workstation space element ID is used as the row and column index of the matrix, and the matrix element values ​​represent the constraint relationships between parts and between parts and workstation space elements, thus forming a workstation structure constraint matrix.

[0018] Specifically, the method for constructing the loading and unloading task queue for components includes:

[0019] Based on the constraint relationships between parts and workstation space elements recorded in the workstation structure constraint matrix, a task dependency table for parts is generated; the task dependency table for parts is transformed into a dependency graph, where nodes represent the loading and unloading tasks of parts, and edges represent the task dependencies that must be executed first; the dependency graph is topologically sorted to obtain a queue of loading and unloading tasks for parts that satisfy all task dependencies.

[0020] Specifically, the method for converting the loading and unloading task queue into a set of action instructions executable by the collaborative robot includes:

[0021] Each operation in the parts loading and unloading task queue is broken down into corresponding action units, which include grab, move, and place operations. For each action unit, the corresponding task sequence, parts identifier, task dependency, and workstation space geometry are bound together.

[0022] Calculate the motion compatibility between each motion unit and the current workstation constraints, define motion unit activation rules based on motion compatibility, determine whether the motion unit is activated, and control the motion activation timing in conjunction with preset workstation cycle requirements;

[0023] The execution of inactive motion units is temporarily suspended until the preset workstation cycle time requirement is met; all activated motion units are combined in the order of the parts loading and unloading task queue to form a set of motion instructions that the collaborative robot can execute.

[0024] Specifically, the preset workstation cycle time requirements include the operation time window, minimum interval between actions, total workstation cycle, and safety buffer time.

[0025] Specifically, the method for generating the candidate path set includes:

[0026] Each action instruction in the action instruction set is represented in three-dimensional space as a path point of the end effector of the collaborative robot for grasping, moving, and placing operations. The end effector path points are projected onto the joint space of the collaborative robot through the robot kinematic model to obtain a joint angle sequence. The joint angles corresponding to each path point are arranged in chronological order to obtain the joint path.

[0027] Based on the constraint information in the workstation structure constraint matrix, the joint path is mapped and corrected so that the collaborative robot end effector avoids occluded areas, maintains the correct stacking order, and meets the preset operation safety gap during movement. For the same set of motion commands, different feasible paths are generated through different joint angle solutions, thereby generating a candidate path set.

[0028] Specifically, the method for selecting the best-performing candidate path as the robot's execution path includes:

[0029] For any candidate path in the candidate path set, the execution time of the collaborative robot to complete the candidate path is obtained by the built-in timer of the collaborative robot controller, and the joint motion energy consumption of the collaborative robot when executing the candidate path is obtained by integrating the angular velocity change of the candidate path.

[0030] Based on the acquired execution time of the collaborative robot in completing the candidate path and the joint motion energy consumption of the collaborative robot when executing the candidate path, a dual-objective optimization function of energy consumption and process cycle time is constructed to optimize and select the candidate path. The candidate path with the best comprehensive scheduling score is selected as the robot's execution path. A cycle time constraint is introduced to ensure that the robot's execution path can be embedded into the production line process cycle time scheduling while satisfying energy efficiency optimization.

[0031] Specifically, the method for inferring collaborative conflicts between collaborative robots and human operations includes:

[0032] By deploying industrial cameras at different angles around the workstation where the collaborative robot performs loading and unloading tasks, the operating area of ​​the collaborative robot's execution path and the area occupied by the manual operation are acquired; bounding box collision detection is then performed on the operating area of ​​the collaborative robot's execution path and the area occupied by the manual operation.

[0033] If the operating area of ​​the collaborative robot's execution path overlaps with the area occupied by the manual operation, it is judged as a collaborative conflict. At the same time, considering the task dependency relationship and the workstation structure constraint matrix, if the corresponding grasping, moving and placing operations of any collaborative robot when executing the action command hinder the manual from completing the loading and unloading tasks, it is also marked as a collaborative conflict.

[0034] Specifically, the method for safely controlling the execution path of the collaborative robot includes:

[0035] Safety regulation is performed on the operation area of ​​the collaborative robot's execution path when there is a collaboration conflict. Safety regulation includes spatial regulation, temporal regulation and path regulation. After safety regulation, the operation area of ​​the collaborative robot's execution path and the area occupied by the human operation are checked again by bounding box collision. If there is still a collaboration conflict, safety regulation is iteratively performed until there is no collaboration conflict.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] This invention combines task dependencies and action compatibility to achieve dynamic matching of action execution sequence and constraints, improving the execution efficiency of collaborative robots in complex workstation environments. Quantifiable and interpretable action compatibility provides a basis for action priority control, parallel action timing constraints, and conservative activation strategies for hazardous components, significantly enhancing system safety. By combining action activation with workstation cycle time requirements, collaborative robots can optimize workstation cycle time while meeting safety and constraint conditions, improving overall production efficiency. Through structured management of action units and workstation constraints, this invention improves the flexibility and adaptability of robot loading and unloading operations, achieving effective integration of macro-level task scheduling and micro-level action control.

[0038] By acquiring the actual execution time of each candidate path and calculating the joint motion energy consumption by integrating the path joint angular velocity, the energy consumption of the trajectory is quantified, reducing the joint motion energy consumption of the collaborative robot and achieving green and energy-saving loading and unloading operations. By introducing constraints such as the upper limit of the workstation cycle time and a safety buffer time, it is ensured that the robot's path selection result will not exceed the maximum allowable cycle time of the production line, enabling loading and unloading tasks to be seamlessly integrated into the production cycle time. This solves the problem that existing technologies cannot adapt to the cycle time of vehicle manufacturing processes. It improves the adaptability of collaborative robots in complex workstations and multi-tasking environments, contributing to improving the automation level and overall efficiency of new energy vehicle manufacturing processes. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the structure of a collaborative robot intelligent loading and unloading system according to the present invention;

[0040] Figure 2 This is a schematic diagram of a collaborative robot intelligent loading and unloading method according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0042] Please see Figure 1 As shown, this embodiment provides a collaborative robot intelligent loading and unloading system, which specifically includes the following steps:

[0043] The workstation structure perception module collects the workstation structure information of the workstation where the collaborative robot is performing loading and unloading tasks, and performs structured encoding on the workstation structure information to generate a workstation structure constraint matrix.

[0044] The component task decomposition module constructs a component loading and unloading task queue based on the workstation structure constraint matrix, and transforms the loading and unloading task queue into a set of action instructions that can be executed by the collaborative robot in combination with the preset workstation cycle requirements.

[0045] The path deduction and avoidance module projects the set of motion commands onto the joint space of the collaborative robot, and generates a set of candidate paths based on the workstation constraint path mapping mechanism. It adopts a dual-objective scheduling of energy consumption and process cycle time to optimize and select the candidate path, and selects the best-performing candidate path as the robot's execution path.

[0046] The collaborative safety control module, when collaborative robots and humans are working at the same workstation, infers collaborative conflicts between the collaborative robots and humans based on the workstation structure constraint matrix and action instruction set, and performs safety control on the execution path of the collaborative robots.

[0047] Workstation structure information includes the location information, posture information, and temporary stacking status information of the components at the workstation.

[0048] The component location information includes three-dimensional coordinate position, bounding box parameters, and relative positional relationships; the attitude information includes the component's rotation angle, quaternion representation, normal vector direction, and symmetry parameters; the temporary stacking status information includes the number of stacking layers, stacking order, contact relationships, available operating space, and stacking stability parameters.

[0049] By deploying 3D vision sensors and depth cameras around the workstation, the components within the workstation are perceived in real time, and 3D point cloud data and image information of the components are obtained. Based on the 3D point cloud data, the components are segmented and identified, and the geometric center coordinates and spatial boundary range of the components are extracted, thereby obtaining the 3D coordinate position, bounding box parameters and relative positional relationship of the components with respect to the workstation reference point in the workstation coordinate system.

[0050] By using point cloud fitting and attitude estimation algorithms, the principal direction vector and rotation angle of the component are calculated, and the rotation angle is converted into a quaternion representation. At the same time, the normal vector direction and symmetry parameters of the main surface of the component are extracted to obtain the attitude information of the component.

[0051] Based on this, the height analysis and contact relationship detection of the overall point cloud of the workstation are performed to determine the level, vertical order and contact boundary of the parts in the stack. Combined with the operable space range around the parts and the center of gravity stability calculation results, the number of stacking layers, stacking order, contact relationship, available operable space and stacking stability parameters of the parts are obtained, thereby obtaining the temporary stacking status information of the parts.

[0052] Methods for generating workstation structure constraint matrices include:

[0053] The collected workstation structure information is standardized. Min-Max normalization is used to convert the component position information into standard three-dimensional coordinates. Euler angle quaternion transformation is used to convert the attitude information into a unified quaternion representation. Temporary stacking status information is converted into quantifiable constraint parameters (temporary stacking status information is converted into numbers or Boolean values ​​(yes / no, 0-1, etc.) to facilitate calculation and automated processing, for example, stacking layer number → value 1, 2, 3, ..., available operating space → volume or radius of the grab area).

[0054] A unique identifier is established for each component, and the component's location information, posture information, and temporary stacking status information are associated with this identifier. At the same time, a geometric code is established for the preset workstation space area to form a workstation space element set.

[0055] Based on the geometric relationships between components and workstation space elements, constraints are generated through mapping. These constraints include adjacency, stacking, occlusion, job priority, and prohibited operations. For example, if part A is above part B, then the loading and unloading tasks of part B must take precedence over those of part A. If part C is occluded by other parts, then the occluding parts must be removed first. The component ID is used as the row index of the matrix, and the workstation space element ID is used as the row and column index. The matrix element values ​​represent the constraints between components and between components and workstation space elements, thus forming the workstation structure constraint matrix.

[0056] Methods for constructing a task queue for loading and unloading parts include:

[0057] Based on the constraint relationships between components and workstation space elements recorded in the workstation structure constraint matrix, a task dependency table for components is generated.

[0058] The task dependency table includes stacking dependencies, occlusion dependencies, spatial conflict dependencies, adjacent dependencies, job priority dependencies, and prohibited operation dependencies. The system scans the workstation structure constraint matrix line by line, and records each "execute first" or "dependency" marker in the task dependency table. For example, if the operation of part A depends on part B to be completed, then A→B is recorded in the task dependency table.

[0059] The task dependency table of components is transformed into a dependency graph, where nodes represent component loading and unloading tasks and edges represent task dependencies that must be executed first. The dependency graph is then topologically sorted to obtain a queue of component loading and unloading tasks that satisfy all task dependencies.

[0060] Methods for converting loading and unloading task queues into a set of action instructions that collaborative robots can execute include:

[0061] Each operation in the parts loading and unloading task queue is broken down into corresponding action units, which include grab, move, and place operations. For each action unit, the corresponding task sequence, parts identifier, task dependency, and workstation space geometry are bound together.

[0062] Calculate the motion compatibility between each motion unit and the current workstation constraints, which include spatial position constraints, occlusion constraints, stacking constraints, dependency constraints, adjacent component constraints, and obstacle constraints; define motion unit activation rules based on motion compatibility to determine whether a motion unit is activated, and control the timing of motion activation in conjunction with preset workstation cycle requirements.

[0063] Action compatibility is: ;in, Indicates action unit At the point of time Motion compatibility with current workstation constraints; Indicates action unit The preset ideal constraint projection vector; The projection vector of the workstation constraint matrix corresponding to the current task queue represents the state projection of the action unit under the current workstation constraint; This represents the weighted L2 norm, used to quantify the motion compatibility between the motion unit and the current workstation constraints; This represents the weight matrix. In calculating the weighted L2 norm, different importance weights are assigned to deviations in different dimensions (because matching motion units with workstation constraints involves multiple dimensions (position, posture, spatial relationships, etc.), and different dimensions may have different degrees of influence on motion compatibility; for example, positional deviation may have a greater impact on task execution success rate than posture deviation). Weighted deviations can be used to make the compatibility calculation more consistent with the current workstation constraints; Indicates the index of the action unit; Index representing a point in time;

[0064] The action unit activation rules are as follows: ;in, Indicates action unit At the point of time The active state; Indicates action unit Activated; Indicates action unit Not activated; This indicates the preset action compatibility threshold;

[0065] The execution of inactive motion units is temporarily suspended until the preset workstation cycle time requirement is met; all activated motion units are combined in the order of the parts loading and unloading task queue to form a set of motion instructions that the collaborative robot can execute.

[0066] The preset workstation cycle time requirements include the operation time window, minimum interval between actions, total workstation cycle, and safety buffer time.

[0067] The operation time window represents the time range within which each part's grasping, moving, and placing actions can be performed at the workstation. It ensures that the robot's actions do not lead or lag, maintaining synchronization with the overall production rhythm of the workstation. The minimum interval between actions represents the minimum safe interval between consecutive actions, ensuring that the robotic arm's movements do not collide with or interfere with manual operations. The total workstation cycle represents the total cycle time for each workstation to complete the loading and unloading operations of a batch of parts, ensuring a consistent rhythm across the entire production line. The safety buffer time represents a safety buffer period set to cope with unforeseen workstation interference or action delays, avoiding conflicts between action commands and manual or other robot operations.

[0068] Methods for generating candidate path sets include:

[0069] Each action command in the action command set, corresponding to the grasping, moving, and placing operations, is represented in three-dimensional space as a path point of the end effector of the collaborative robot. The end effector path points are projected onto the joint space of the collaborative robot using the robot kinematics model (including forward kinematics model and inverse kinematics model) to obtain the joint angle sequence (given the spatial position and orientation of the end effector of the path point, the joint angle that enables the end effector to reach the position is solved by the inverse kinematics model). The joint angles corresponding to each path point are arranged in chronological order to obtain the joint path.

[0070] It should be noted that each action command corresponds to a spatial operation, such as grasping a part, moving to a certain position, or placing a part. The collaborative robot's end effector (gripper or tool) has position and orientation information in the three-dimensional workspace. For grasping operations: end effector path point = position of the part grasping point + corresponding grasping orientation (Eulerian angles or quaternions); for movement operations: the continuous path is discretized into multiple path points, each path point recording the end effector's three-dimensional coordinates and orientation at that position; for placement operations: end effector path point = placement position + placement orientation; thus, grasping, movement, and placement operations are transformed into a sequence of path points that can be represented in three-dimensional space, with each path point representing the spatial state of the end effector at that point in time.

[0071] Based on the constraint information in the workstation structure constraint matrix, the joint path is mapped and corrected so that the collaborative robot end effector avoids occlusion areas, maintains the correct stacking order, and meets the preset operation safety gap during movement. For the same set of motion commands, different feasible paths are generated through different joint angle solutions (for the same end effector path point, the inverse kinematics model may generate multiple joint angle solutions (different joint angle combinations), each of which can allow the collaborative robot end effector to reach the target position), thereby generating a candidate path set.

[0072] Methods for selecting the best-performing candidate path as the robot's execution path include:

[0073] For any candidate path in the candidate path set, the execution time of the collaborative robot to complete the candidate path is obtained by the built-in timer of the collaborative robot controller, and the joint motion energy consumption of the collaborative robot when executing the candidate path is obtained by integrating the angular velocity change of the candidate path.

[0074] The energy consumption of joint movement is: ;in, This indicates that the collaborative robot is executing candidate paths. Energy consumption during joint movement; Represents any candidate path in the candidate path set; This represents the joint angle sequence of the candidate path in the normalized interval [0,1]. This represents the joint angular velocity vector of the candidate path within the normalized interval; The parameter represents the normalized interval [0,1], used to represent the progress percentage of the candidate path from the start point to the end point, for example... Indicates the starting point of the candidate path;

[0075] Based on the acquired execution time of the collaborative robot in completing the candidate path and the joint motion energy consumption of the collaborative robot when executing the candidate path, a dual-objective optimization function of energy consumption and process cycle time is constructed to optimize and select the candidate path. The candidate path with the best comprehensive scheduling score is selected as the robot's execution path.

[0076] The bi-objective optimization function is: ;in, Indicate candidate path The comprehensive scheduling score is used to simultaneously measure the execution time and energy consumption of candidate paths; This indicates that the collaborative robot has completed the candidate path. Execution time; This represents the candidate path with the longest execution time in the candidate path set. This represents the candidate path with the highest joint motion energy consumption in the candidate path set; it represents the scheduling weight factor, used to balance execution time and energy consumption.

[0077] Cycle time constraints are introduced to ensure that the robot's execution path can be embedded into the production line's process cycle time scheduling while meeting energy efficiency optimization requirements. The cycle time constraints are as follows: ; where represents the upper limit of the workstation cycle time, that is, the maximum allowable time for the collaborative robot to complete the loading and unloading tasks required by the production line; and represents the preset safety buffer time, which is used to avoid cycle imbalance or task delay and to ensure that the path execution does not exceed the production line cycle.

[0078] Methods for inferring collaborative conflicts between collaborative robots and human tasks include:

[0079] By deploying industrial cameras at different angles around the workstation where the collaborative robot performs loading and unloading tasks, the operating area of ​​the collaborative robot's execution path and the area occupied by the manual operation are acquired; bounding box collision detection is then performed on the operating area of ​​the collaborative robot's execution path and the area occupied by the manual operation.

[0080] If the operating area of ​​the collaborative robot's execution path overlaps with the area occupied by the manual operation, it is judged as a collaborative conflict. At the same time, considering the task dependency relationship and the workstation structure constraint matrix, if the corresponding grasping, moving and placing operations of any collaborative robot when executing the action command hinder the manual from completing the loading and unloading tasks, it is also marked as a collaborative conflict.

[0081] Logical obstacle assessment based on task dependencies: If the robot's operation alters or blocks the parts the human needs to manipulate (e.g., the robot places a part, preventing the human from grasping other parts), check if the part or area corresponding to the robot's current action is a task node on which the human's current operation depends. If so, mark it as a potential obstacle.

[0082] Structural obstruction identification based on workstation structure constraint matrix: Robot actions prevent humans from completing operations according to the spatial rules of the constraint matrix. For example, the robot occupies the operating surface, tool path, or part stacking position that the human must use. Map the space occupied by the robot's actions to the workstation structure matrix to check whether the human operation path or critical components are blocked.

[0083] Methods for safely controlling the execution path of collaborative robots include:

[0084] Safety regulation is performed on the operation area of ​​the collaborative robot's execution path when there is a collaboration conflict. Safety regulation includes spatial regulation, temporal regulation and path regulation. After safety regulation, the operation area of ​​the collaborative robot's execution path and the area occupied by the human operation are checked again by bounding box collision. If there is still a collaboration conflict, safety regulation is iteratively performed until there is no collaboration conflict.

[0085] Spatial control involves offsetting or fine-tuning in three-dimensional space to avoid areas occupied by humans, while maintaining the task stacking order, dependencies, and workstation constraints. This is achieved by modifying the end effector trajectory points to ensure smooth completion of the motion operation. Temporal control involves adjusting the execution timing of motion units to achieve safe isolation. This includes increasing delays or changing the minimum interval between actions, ensuring that robot and human operations at the same location do not occur simultaneously, while meeting preset workstation cycle time requirements and allowing for a safety buffer. Path control involves replanning the entire execution path, generating multiple feasible paths, and using path planning algorithms to select the trajectory that satisfies safety constraints and is most efficient. This considers robot joint spatial motion limitations, velocity, and acceleration constraints to ensure a smooth and executable trajectory.

[0086] The preset action compatibility threshold is set by the staff. By collecting different action compatibility data, the average value of multiple action compatibility data is taken as the preset action compatibility threshold. It can be adjusted by the staff during system operation according to the actual situation.

[0087] This embodiment achieves dynamic matching of action execution sequence and constraints by combining task dependencies and action compatibility, improving the execution efficiency of collaborative robots in complex workstation environments. Quantifiable and interpretable action compatibility provides a basis for action priority control, parallel action timing constraints, and conservative activation strategies for hazardous components, significantly enhancing system safety. Combining action activation with workstation cycle time requirements allows collaborative robots to optimize workstation cycle time while meeting safety and constraint conditions, improving overall production efficiency. Through structured management of action units and workstation constraints, this invention improves the flexibility and adaptability of robot loading and unloading operations, achieving effective integration of macro-level task scheduling and micro-level action control.

[0088] By acquiring the actual execution time of each candidate path and calculating the joint motion energy consumption by integrating the path joint angular velocity, the energy consumption of the trajectory is quantified, reducing the joint motion energy consumption of the collaborative robot and achieving green and energy-saving loading and unloading operations. By introducing constraints such as the upper limit of the workstation cycle time and a safety buffer time, it is ensured that the robot's path selection result will not exceed the maximum allowable cycle time of the production line, enabling loading and unloading tasks to be seamlessly integrated into the production cycle time. This solves the problem that existing technologies cannot adapt to the cycle time of vehicle manufacturing processes. It improves the adaptability of collaborative robots in complex workstations and multi-tasking environments, contributing to improving the automation level and overall efficiency of new energy vehicle manufacturing processes. Example

[0089] Please see Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A collaborative robot intelligent loading and unloading method is provided, including:

[0090] S1. Collect the workstation structure information of the workstation where the collaborative robot is performing loading and unloading tasks, and perform structured encoding on the workstation structure information to generate a workstation structure constraint matrix.

[0091] S2. Based on the workstation structure constraint matrix, construct the loading and unloading task queue of parts, and combine it with the preset workstation cycle requirements to transform the loading and unloading task queue into a set of action instructions that can be executed by the collaborative robot.

[0092] S3. Project the set of motion instructions onto the joint space of the collaborative robot, and generate a set of candidate paths based on the workstation constraint path mapping mechanism; use dual-objective scheduling of energy consumption and process cycle time to optimize the selection of candidate paths, and select the candidate path with the best performance as the robot's execution path.

[0093] S4. When collaborative robots and humans work at the same workstation, based on the workstation structure constraint matrix and action instruction set, infer the collaborative conflicts between the collaborative robots and humans, and perform safety control on the execution path of the collaborative robots.

[0094] Since the electronic device described in this embodiment is the electronic device used in implementing the intelligent loading and unloading system for collaborative robots in this application embodiment, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the intelligent loading and unloading system for collaborative robots described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. As long as those skilled in the art implement the electronic device used in the intelligent loading and unloading system for collaborative robots in this application embodiment, it falls within the protection scope of this application.

[0095] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0096] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A collaborative robot intelligent loading and unloading system, characterized in that, include: The workstation structure perception module collects the workstation structure information of the workstation where the collaborative robot is performing loading and unloading tasks, and performs structured encoding on the workstation structure information to generate a workstation structure constraint matrix. The workstation structure information includes the component location information, posture information, and temporary stacking status information of the workstation; The method for generating the workstation structure constraint matrix includes: The collected workstation structure information is standardized. The component position information is transformed into standard three-dimensional coordinates through Min-Max normalization. The attitude information is transformed into a unified quaternion representation using Euler angle quaternion transformation. The temporary stacking state information is transformed into quantifiable constraint parameters. A unique identifier is established for each component, and the component's location information, posture information, and temporary stacking status information are associated with this identifier. At the same time, a geometric code is established for the preset workstation space area to form a workstation space element set. Based on the geometric relationship between the parts and the workstation space element set, constraint relationships are generated by mapping. The part ID is used as the row index of the matrix, the workstation space element ID is used as the row and column index of the matrix, and the matrix element values ​​represent the constraint relationships between parts and between parts and workstation space elements, thus forming a workstation structure constraint matrix. The component task decomposition module constructs a component loading and unloading task queue based on the workstation structure constraint matrix, and transforms the loading and unloading task queue into a set of action instructions that can be executed by the collaborative robot in combination with the preset workstation cycle requirements. The method for constructing the loading and unloading task queue for components includes: Based on the constraint relationships between parts and workstation space elements recorded in the workstation structure constraint matrix, a task dependency table for parts is generated; the task dependency table for parts is transformed into a dependency graph, where nodes represent the loading and unloading tasks of parts, and edges represent the task dependencies that must be executed first; the dependency graph is topologically sorted to obtain a queue of loading and unloading tasks for parts that satisfy all task dependencies. The path deduction and avoidance module projects the set of motion commands onto the joint space of the collaborative robot, and generates a set of candidate paths based on the workstation constraint path mapping mechanism. It adopts a dual-objective scheduling of energy consumption and process cycle time to optimize and select the candidate path, and selects the candidate path with the best performance as the robot's execution path. The collaborative safety control module, when collaborative robots and humans are working at the same workstation, infers collaborative conflicts between the collaborative robots and humans based on the workstation structure constraint matrix and action instruction set, and performs safety control on the execution path of the collaborative robots.

2. The intelligent loading and unloading system for collaborative robots according to claim 1, characterized in that, The method for converting the loading and unloading task queue into a set of action instructions that can be executed by the collaborative robot includes: Each operation in the parts loading and unloading task queue is broken down into corresponding action units, which include grab, move, and place operations. For each action unit, the corresponding task sequence, parts identifier, task dependency relationship, and workstation space geometric relationship are bound. Calculate the motion compatibility between each motion unit and the current workstation constraints, define motion unit activation rules based on motion compatibility, determine whether the motion unit is activated, and control the motion activation timing in conjunction with preset workstation cycle requirements; The execution of inactive motion units is temporarily suspended until the preset workstation cycle time requirement is met; all activated motion units are combined in the order of the parts loading and unloading task queue to form a set of motion instructions that the collaborative robot can execute.

3. The intelligent loading and unloading system for collaborative robots according to claim 2, characterized in that, The preset workstation cycle time requirements include the operation time window, minimum interval between actions, total workstation cycle, and safety buffer time.

4. The intelligent loading and unloading system for collaborative robots according to claim 3, characterized in that, The method for generating the candidate path set includes: Each action command in the motion command set, corresponding to the grasping, moving, and placing operations, is represented in 3D space as a path point of the collaborative robot's end effector. Using the robot's kinematic model, these path points are projected onto the collaborative robot's joint space to obtain a sequence of joint angles. The joint angles corresponding to each path point are then arranged in chronological order to obtain the joint path. Based on the constraint information in the workstation structure constraint matrix, the joint path is mapped and corrected to ensure that the collaborative robot's end effector avoids occlusion areas, maintains the correct stacking order, and meets preset operational safety clearances during movement. For the same motion command set, different feasible paths are generated through different joint angle solutions, thus generating a candidate path set.

5. The intelligent loading and unloading system for collaborative robots according to claim 4, characterized in that, The method for selecting the best-performing candidate path as the robot's execution path includes: For any candidate path in the candidate path set, the execution time of the collaborative robot to complete the candidate path is obtained by the built-in timer of the collaborative robot controller, and the joint motion energy consumption of the collaborative robot when executing the candidate path is obtained by integrating the angular velocity change of the candidate path. Based on the acquired execution time of the collaborative robot in completing the candidate path and the joint motion energy consumption of the collaborative robot when executing the candidate path, a dual-objective optimization function of energy consumption and process cycle time is constructed to optimize and select the candidate path. The candidate path with the best comprehensive scheduling score is selected as the robot's execution path. A cycle time constraint is introduced to ensure that the robot's execution path can be embedded into the production line process cycle time scheduling while satisfying energy efficiency optimization.

6. The intelligent loading and unloading system for collaborative robots according to claim 5, characterized in that, The methods for inferring collaborative conflicts between collaborative robots and human operations include: By deploying industrial cameras at different angles around the workstation where the collaborative robot performs loading and unloading tasks, the operating area of ​​the collaborative robot's execution path and the area occupied by the manual operation are acquired; bounding box collision detection is then performed on the operating area of ​​the collaborative robot's execution path and the area occupied by the manual operation. If the operating area of ​​the collaborative robot's execution path overlaps with the area occupied by the manual operation, it is judged as a collaborative conflict. At the same time, considering the task dependency relationship and the workstation structure constraint matrix, if the corresponding grasping, moving and placing operations of any collaborative robot when executing the action command hinder the manual from completing the loading and unloading tasks, it is also marked as a collaborative conflict.

7. The intelligent loading and unloading system for collaborative robots according to claim 6, characterized in that, The method for safely controlling the execution path of collaborative robots includes: Safety regulation is performed on the operation area of ​​the collaborative robot's execution path when there is a collaboration conflict. Safety regulation includes spatial regulation, temporal regulation and path regulation. After safety regulation, the operation area of ​​the collaborative robot's execution path and the area occupied by the human operation are checked again by bounding box collision. If there is still a collaboration conflict, safety regulation is iteratively performed until there is no collaboration conflict.

Images