Dual-arm compound robot and its control system

Abstract

The invention discloses a dual-arm composite robot and a control system thereof, and relates to the technical field of dual-arm composite robots. After a target position is determined by a task to be executed, a target path for the robot to move to the target position is planned, dynamically updated and marked on an electronic map; after the collision risk of the robot is predicted and the action state of the robot is adaptively adjusted, at least the end of the robot arm is made to reach the target position; after automatically switching the assembly task, corresponding feedback data is collected, if the feedback value generated by the feedback data does not exceed expectations, the current assembly task switching mode is switched, a collision value is generated by the recorded robot collision risk, and when the collision value exceeds expectations, a risk area is set in the assembly area, and constraints are imposed on the adjustment frequency of the risk area; and by automatically switching the task allocation mode, the efficiency of executing the assembly task is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of dual-arm compound robots, in particular to a dual-arm compound robot and a control system thereof. Background Art

[0002] Dual-arm robots are highly advanced robotic systems with two independent manipulators that can mimic the collaborative principles of human arms and achieve a variety of complex movements and operations. They are usually equipped with advanced sensors, controllers, and actuators that can accurately sense the environment, understand tasks, and complete tasks efficiently. Dual-arm robots are highly flexible and safe and can operate in a variety of complex environments, such as industrial automation production lines, emergency rescue sites, and educational and scientific research laboratories. They can complete various tasks autonomously or in collaboration with humans, such as material handling, assembly, processing, and testing, greatly improving production efficiency and operational safety.

[0003] In the Chinese invention patent with application publication number CN114670202A, a control system of a ball balancing robot is disclosed, wherein the ball balancing robot includes a ball balancing mechanism, an operating arm mechanism, a main controller and a camera, the operating arm mechanism is provided with two mechanical arms, and the control system of the ball balancing robot includes the following operation modes: a single-arm operation mode, in which one mechanical arm and a ball wheel jointly support the ground, and the other mechanical arm performs a space operation, a climbing mode, in which the two mechanical arms support the ground so that the ball wheel is separated from the ground and climbs over obstacles, and a double-arm operation mode, in which the ball wheel supports the ground and the two mechanical arms perform a space operation, and the main controller controls the posture of the ball balancing mechanism, the posture of the operating arm mechanism and the posture of the steering wheel according to the acquired data of the camera to execute one of the operation modes. According to the control system of the ball balancing robot of the present invention, different motion modes can be adopted, and the motion mode is flexible, so that mobile operation and obstacle crossing can be realized, and environmental applicability is improved.

[0004] When a dual-arm robot, especially a mobile dual-arm robot, is assembling assembly parts, considering the diversity and complexity of the assembly tasks, it is necessary to constantly adjust the position of the robot and the state of the robotic arm; when there are many assembly tasks, the existing robot control methods usually sort the various assembly tasks according to their content and importance, and then execute them in sequence.

[0005] However, on the intelligent manufacturing assembly line, the mobile dual-arm robot operates in a limited workspace and needs to handle multiple tasks at the same time, such as picking up parts from the conveyor belt and assembling multiple complex parts. When there are assembly workers collaborating on the assembly line, if the assembly tasks are still performed sequentially according to the pre-built assembly task list, the dual-arm robot and its robotic arm may produce more repetitive and invalid movements. At the same time, when the robot and the robotic arm move frequently, it is more likely to collide, and even collide with the assembly workers, which not only affects the assembly efficiency but also poses certain safety hazards.

[0006] To this end, the present invention provides a dual-arm compound robot and a control system thereof. Summary of the invention

[0007] 1. Technical issues to be resolved

[0008] In view of the deficiencies in the prior art, the present invention provides a dual-arm compound robot and a control system thereof. After determining the target position by the task to be executed, the target path for the robot to move to the target position is planned, dynamically updated and marked on an electronic map; after predicting the collision risk of the robot and adaptively adjusting the action state of the robot, at least the end of the robot arm reaches the target position; after automatically switching the assembly task, corresponding feedback data is collected. If the feedback value generated by the feedback data does not exceed expectations, the current assembly task switching mode is switched, a collision value is generated by the recorded robot collision risk, and when the collision value exceeds expectations, a risk area is set in the assembly area, and constraints are imposed on the adjustment frequency of the risk area; by automatically switching the task allocation mode, the efficiency of executing the assembly task is improved, thereby solving the technical problems recorded in the background technology.

[0009] (II) Technical solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: a control system of a dual-arm compound robot includes a task scheduling module, which sorts assembly tasks according to priority, and if the assembly load does not exceed expectations, matches the tasks to be executed from the high-priority assembly tasks according to the current posture of the robot arm;

[0011] The path planning module plans the robot to move to the target location after the target location is determined by the task to be executed, and dynamically updates and marks it on the electronic map;

[0012] The motion control module predicts the robot's collision risk and adaptively adjusts the robot's motion state to at least make the end of the robot arm reach the target position;

[0013] A mode control module collects corresponding feedback data after automatically switching the assembly task, and switches the current assembly task switching mode if the feedback value generated by the feedback data does not exceed the expectation;

[0014] The safety control module generates a collision value based on the recorded robot collision risk. When the collision value exceeds the expected value, a risk area is set in the assembly area and the adjustment frequency of the risk area is constrained. The way to constrain the adjustment frequency of the risk area is as follows:

[0015] ;

[0016] Where: is the assembly cycle, is the distance threshold, The value is 2.718, is the time attenuation coefficient, ranging from 0 to 1. is the distance attenuation coefficient, ranging from 0.01 to 0.1; is the first j Alarm nodes, For the i and j The time interval between alarm nodes, is the number of alarm nodes, For the j The distance into the risk area on each alarm node.

[0017] Furthermore, the current assembly task description data is collected and used as input. The trained task scoring model is used to output the corresponding task score. After marking the assembly tasks with the task scores, the assembly tasks are sorted from high to low and a task list is generated.

[0018] Furthermore, the assembly task data in the current assembly cycle is collected, and the task density of the assembly tasks in the current assembly cycle is generated from the assembly task data set. If the task density does not exceed the density threshold, a new task addition permission is issued to the outside.

[0019] Furthermore, after obtaining permission to add new tasks, task priorities are dynamically adjusted when new tasks are inserted;

[0020] After obtaining several assembly tasks with higher task priorities in the task list, the robot arm's running trajectory is estimated when executing each corresponding assembly task; the work task with the highest degree of matching with the robot arm's current state and running trajectory is selected as the task to be executed.

[0021] Furthermore, image data in the assembly area is acquired in real time, target detection and recognition is performed on the acquired image data, the type, volume and position data of the assembly parts in the assembly area are acquired, and the position data of obstacles and assembly workers are detected;

[0022] Construct an electronic map covering the assembly area and mark the acquired location data on the electronic map.

[0023] Furthermore, after the task to be executed is selected, the execution position of the task to be executed is used as the target position, and the shortest path from the starting position to the target position is planned for the robot by the Dijkstra algorithm on the electronic map;

[0024] The electronic map is updated in real time, the shortest path is adjusted according to the updated information, and the adjusted shortest path is used as the target path and marked on the electronic map.

[0025] Furthermore, obstacle data in the assembly area is collected, and the obstacle data is used as input to predict the collision risk of the robot using the trained collision prediction model, and the corresponding collision risk value is output;

[0026] When the collision risk value exceeds the preset risk threshold, the action of the robotic arm and the movement of the robot are stopped.

[0027] Furthermore, the location of the assembly to be assembled or the expected arrival position is used as the target position of the end of the robot arm, the inverse kinematics algorithm is used to calculate the angle and range of motion of each joint, and a PID controller is applied to each joint to adjust the motion trajectory of the robot arm in real time so that the end of the robot arm reaches the target position.

[0028] Furthermore, the position and motion data of the assembly workers are monitored and collected in real time, and the real-time motion data of the robot arm and the current motion trajectory data are recorded and summarized as task status data; the assembly task with the highest degree of match with the current task status data is selected at the front of the task list as the task to be executed.

[0029] Furthermore, feedback data is collected when tasks are automatically switched, and a feedback value is constructed from the feedback data. If the feedback value does not exceed a preset feedback threshold, the automatic mode switching mechanism is triggered to switch the automatic control mode of the current assembly task to an active control mode. The microphone receives the worker's voice commands, and the worker's commands are converted into control commands through semantic recognition and extraction.

[0030] Furthermore, in the active control mode, the collision risk value of each sub-cycle is continuously collected, and the collision value is generated from the continuous collision risk values. If the acquired collision value exceeds the collision threshold, a risk area is set in the assembly area with the robot position as the center. The area of ​​the risk area is proportional to the ratio of the collision value exceeding the collision threshold, and the risk area is dynamically adjusted in real time.

[0031] When the robot detects the approach of an assembler or an obstacle, it will automatically slow down; when the assembler enters the risk area, the robot's movement will automatically slow down or stop, the robotic arm will pause and send an alarm to the outside.

[0032] A dual-arm compound robot comprises at least an upper arm and a lower arm, a motion joint providing multiple degrees of freedom, an end effector, a power unit for driving the joint and the arm movement; a sensor network; and at least a position sensor, a torque sensor, and a visual sensor.

[0033] (III) Beneficial effects

[0034] The present invention provides a dual-arm compound robot and a control system thereof, which have the following beneficial effects:

[0035] 1. Judge the current assembly load level based on the task density to prevent the robot from being in a high-load operation state for a long time; select the task that best matches the current state of the robot arm from the first several assembly tasks as the task to be executed, which can reduce the current activity of the robot arm and improve the execution efficiency of the assembly task.

[0036] 2. Planning the corresponding moving path for the robot and updating the moving path in real time can improve the robot's moving efficiency and prevent the robot from colliding with obstacles and assembly workers in the assembly area while in motion; predicting the collision risk through the collision prediction model, and performing emergency control on the robot and its robotic arm when there is a collision risk or the collision risk exceeds expectations.

[0037] 3. Select a new assembly task from among several assembly tasks with higher priorities, and automatically switch the assembly tasks based on the analysis of the priorities of each assembly task, thereby improving the efficiency of executing the assembly tasks.

[0038] 4. Judge and evaluate the execution effect of the assembly task under the current control mode based on the feedback value, automatically switch the task switching mode when the feedback effect changes, and automatically switch the task allocation mode to adapt to the current assembly task scenario.

[0039] 5. Determine the current collision risk based on the collision value. If the overall collision risk is high, it is necessary to set up a risk area in the vicinity of the robot and the robotic arm to ensure the safety of the assembly workers.

[0040] 6. By adjusting the size of the risk area in real time, limiting obstacles and the activity area of ​​​​assembly workers, and sending alarms to the outside in a timely manner, different adjustment methods can be adopted for robots and robotic arms to reduce the occurrence of collision incidents and reduce the operation risk of robotic arms; the adjustment frequency is constrained according to the frequency of assembly workers entering the risk area and the corresponding position, so that the switching adjustment of the risk area is adapted to the changes in the actual assembly tasks, thus ensuring assembly safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the control method of the dual-arm compound robot of the present invention;

[0042] Figure 2 This is a schematic diagram of the control system structure of the dual-arm compound robot of the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] See also Figure 1 The present invention provides a control method for a dual-arm compound robot, comprising:

[0045] Step 1: Sort the assembly tasks according to priority. If the assembly load does not exceed expectations, match the tasks to be executed from the high-priority assembly tasks according to the current posture of the robot arm.

[0046] The step 1 includes the following contents:

[0047] Step 101: When a dual-arm robot is required to assemble the assembly parts in the assembly area, current assembly task description data is collected, such as the content of the assembly task, the time sensitivity of the assembly task, the degree of impact on the overall production process, the occupation of required resources, whether it depends on the completion of other tasks, etc.;

[0048] After scoring and labeling the description data, the convolutional neural network is trained with the labeled sample data to obtain the trained task scoring model; the description data of the assembly task is used as input, and the trained task scoring model is used to output the corresponding task score. After labeling the assembly tasks with the task score, the assembly tasks are sorted from high to low and a task list is generated;

[0049] When in use, when there are multiple assembly tasks, the content and importance of each assembly task are determined, and each assembly task is sorted to determine the priority of executing each assembly task. Important tasks are executed first, thereby improving the overall efficiency of task execution.

[0050] Step 102: Collect the assembly task data in the current assembly cycle, including the execution location data and the estimated completion time of the assembly task, and generate an assembly task data set after summarizing; under dimensionless conditions, generate the task density of the assembly task in the current assembly cycle from the assembly task data set. , as follows:

[0051] ;

[0052] Weight coefficient: , , It is i Assembly tasks to j The time interval between assembly tasks, is the mean of the time interval, It is i Assembly tasks to j The position interval of each assembly task, is the mean of the position intervals;

[0053] Based on historical data and management expectations for assembly tasks, density thresholds are pre-set; if the task density is If the density threshold is exceeded, it means that the current task scheduling state is close to saturation and new assembly tasks cannot be inserted. Otherwise, a new task addition permission is issued to the outside.

[0054] When used, after there are several assembly tasks in the task list, the task density is pre-built based on the expected execution data and distribution data of each assembly task. , depending on the task intensity The density of the current assembly tasks and the degree of assembly load are judged. If the load level is high, no new assembly tasks are added, which can prevent the robot from being in a high-load operation state for a long time.

[0055] Step 103: After obtaining permission to add a new task, dynamically adjust the task priority when a new task is inserted, and generate a new task list; after obtaining several assembly tasks with higher task priority in the task list, estimate the robot arm's running trajectory when executing each corresponding assembly task; select the work task with the highest degree of matching with the current state and running trajectory of the robot arm as the task to be executed;

[0056] When using, combine the contents in steps 101 to 103:

[0057] Taking the current state of the robot's manipulator as the starting point, the task that best matches the current state of the manipulator is selected from the first several assembly tasks as the task to be executed. This can improve the execution efficiency of the assembly task by reducing the current activity of the manipulator.

[0058] However, on the intelligent manufacturing assembly line, the mobile dual-arm robot operates in a limited workspace and needs to handle multiple tasks at the same time, such as picking up parts from the conveyor belt and assembling multiple complex parts. When there are assembly workers collaborating on the assembly line, if the assembly tasks are still performed sequentially according to the pre-built assembly task list, the dual-arm robot and its robotic arm may produce more repetitive and invalid movements. At the same time, when the robot and the robotic arm move frequently, it is more likely to collide, and even collide with the assembly workers, which not only affects the assembly efficiency but also poses certain safety hazards.

[0059] Step 2: After the target location is determined by the task to be executed, the target path for the robot to move to the target location is planned, dynamically updated and marked on the electronic map;

[0060] The step 2 includes the following contents:

[0061] Step 201: The image acquisition device acquires image data in the assembly area in real time, performs target detection and recognition on the acquired image data, acquires relevant data such as the type, volume and position of the assembly parts in the assembly area, and detects obstacles in the surrounding environment and the position data of the assembly workers in combination with the distance sensor;

[0062] Step 202: construct an electronic map covering the assembly area, mark the acquired position data on the electronic map, select the task to be executed, take the execution position of the task to be executed as the target position, and plan the shortest path from the starting position to the target position for the robot on the electronic map by using the Dijkstra algorithm;

[0063] Update the electronic map in real time, adjust the shortest path according to the updated information, and mark the adjusted shortest path as the target path on the electronic map;

[0064] When using, combine the contents in steps 201 and 202:

[0065] After selecting the assembly task to be performed, when the robot can move, a corresponding moving path is planned for the robot, and the moving path is updated in real time. By planning the shortest path, the robot's movement efficiency can be improved. At the same time, it can also prevent the robot from colliding with obstacles and assembly workers in the assembly area while in motion.

[0066] Step 3: predict the robot's collision risk and adaptively adjust the robot's motion state to at least make the end of the robot arm reach the target position;

[0067] The step three includes the following contents:

[0068] Step 301: train a convolutional neural network with the labeled sample data to obtain a trained collision prediction model; collect motion data, position data, and distance data of each mobile obstacle and assembly worker in the assembly area through the sensor network, and use them as obstacle data; use the obstacle data as input, use the trained collision prediction model to predict the possible collision risk of the robot, and output the corresponding collision risk value. When the collision risk value exceeds the preset risk threshold, stop the action of the robot arm and the movement of the robot;

[0069] Step 302: The position of the assembly to be assembled or the expected arrival position is used as the target position of the end of the robot arm, the inverse kinematics algorithm is used to calculate the angle and range of motion of each joint, and a PID controller is applied to each joint to adjust the motion trajectory of the robot arm in real time so that the end of the robot arm reaches the target position;

[0070] When used, combine the contents in steps 301 and 302:

[0071] When the robot is in a moving state, the collision risk is predicted through the collision prediction model. When there is a collision risk or the collision risk exceeds expectations, emergency control is performed on the robot and its robotic arm, such as adjusting the motion trajectory of the robot or the robotic arm to adjust the current collision risk. After calculating the motion trajectory of the robotic arm, the output motion trajectory of the robotic arm can be simulated and tested, and the motion trajectory of the robotic arm can be optimized based on the verified test results, so that the robot and its robotic arm can perform assembly tasks with higher reliability and accuracy.

[0072] Step 4: After automatically switching the assembly task, collect corresponding feedback data, and if the feedback value generated by the feedback data does not exceed expectations, adjust the current assembly task switching mode;

[0073] The step 4 includes the following contents:

[0074] Step 401: When the robot performs an assembly task in the task list, the position and motion data of the assembly worker are collected in real time through the visual sensor and the distance sensor, and the real-time motion data of the robot arm and the current motion trajectory data are recorded, and the data are summarized as the task status data;

[0075] After the current assembly task is completed, the assembly task with the highest matching degree of the current task status data is selected from the front column of the task list as the task to be executed;

[0076] When in use, after the previous assembly task is completed, a new assembly task is selected from several assembly tasks with higher priorities based on the current posture and position of the robot arm. Based on the analysis and acquisition of the priority of each assembly task, automatic switching of assembly tasks is achieved to improve the efficiency of executing assembly tasks.

[0077] Step 402: Collect feedback data when the task is automatically switched, including the execution time of all assembly tasks, the position change data of the robot and the energy consumption data, etc., and aggregate the acquired feedback data to generate a feedback data set;

[0078] Under dimensionless conditions, the feedback value is constructed from the feedback data in the feedback data set. , as follows:

[0079] ;

[0080] Where: and are weight coefficients respectively, the sum of the two is 1, and the values ​​fall into , the value is consistent with the previous value, To perform all assembly tasks, Position change distance; is the mean of the shortest distances, is the energy consumption of the robot at time t, is the qualified value of energy consumption, is its mean value, is the total energy consumption;

[0081] If the feedback value If the preset feedback threshold is not exceeded, it means that the feedback effect of the automatic switching of the current assembly task has not met expectations. If the existing assembly task is still automatically switched, the execution efficiency of the current assembly task may not be improved. At this time, the automatic mode switching mechanism is triggered to switch the automatic control mode of the current assembly task to the active control mode. In the active control mode, the microphone receives the worker's voice command, and the worker's command is converted into a control command through semantic recognition and extraction.

[0082] When using, combine the contents in steps 401 and 402:

[0083] When the assembly task is in active switching mode, the feedback data in the current control mode is collected within the preset observation period, and the feedback value is constructed from the feedback data. , according to the feedback value Judge and evaluate the execution effect of the assembly task under the current control mode, automatically switch the task switching mode when the feedback effect changes, and adapt to the current assembly task scenario by automatically switching the task allocation mode.

[0084] Step 5: Generate collision value from recorded robot collision risk , at the collision value When it exceeds expectations, a risk area is set up in the assembly area, and restrictions are imposed on the adjustment frequency of the risk area;

[0085] The step five includes the following contents:

[0086] Step 501: Continuously collect the collision risk values ​​of each sub-period in the active control mode, and generate the collision value from the continuous collision risk values ​​under dimensionless conditions. , as follows:

[0087] ;

[0088] in, , is the number of sub-periods, and the weight coefficient is: , and ; For the i The collision risk value within a sub-period is is the mean value of collision risk, is the acceptable value of the collision risk value.

[0089] Pre-set collision thresholds based on historical data and expectations for collision risk management;

[0090] If you get the collision value If the collision threshold is exceeded, it means that in the active control mode, the robot and its mechanical arm are at a high risk of colliding with obstacles or assembly workers, and the risk area opening mechanism needs to be activated in the assembly area. At this time, a start command in the risk area is issued to the outside;

[0091] When in use, the collision risk of the current assembly task is recorded during execution, and a collision value is constructed from continuous collision risks. The current collision risk is judged based on the collision value. If the overall collision risk level is high, it is necessary to set up a risk area in the vicinity of the robot and the robotic arm to ensure the safety of the assembly workers.

[0092] Step 502: After receiving the start command, a risk area is set in the assembly area with the robot position as the center. The area of ​​the risk area is proportional to the ratio of the collision value exceeding the collision threshold, and the risk area is dynamically adjusted in real time based on the current assembly task and the position of the robot.

[0093] When an assembly worker or obstacle is detected approaching, the robot will automatically slow down; when an assembly worker enters a risk area, the robot will automatically slow down or stop, the robot arm will pause, and an alarm will be sent to the outside.

[0094] When in use, by adjusting the size of the risk area in real time, limiting obstacles and the activity area of ​​​​assembly workers, and promptly sending alarms to the outside, and taking different adjustment methods for robots and robotic arms, the occurrence of collision incidents can be reduced and the operation risk of the robotic arm can be reduced.

[0095] Step 503: In order to prevent the risk area from being frequently updated and switched, the adjustment frequency of the risk area is constrained under dimensionless conditions so that the adjustment frequency in the risk area within the preset assembly cycle is within a controllable range, as follows:

[0096] ;

[0097] Where: is the assembly cycle, is the distance threshold, The value is 2.718, is the time attenuation coefficient, ranging from 0 to 1. is the distance attenuation coefficient, ranging from 0.01 to 0.1; is the first j Alarm nodes, For the i and j The time interval between alarm nodes, is the number of alarm nodes, For the j The distance within the risk area of ​​each alarm node;

[0098] When using, combine the contents in steps 501 to 503:

[0099] Taking into account that the risk area should not be adjusted too frequently, the adjustment frequency is constrained according to the frequency of assembly workers entering the risk area and the corresponding position, so that the switching adjustment of the risk area is adapted to the changes in the actual assembly tasks, further improving the guarantee of assembly safety.

[0100] See also Figure 2 The present invention provides a control system for a dual-arm compound robot, comprising:

[0101] The task scheduling module sorts the assembly tasks according to their priorities. If the assembly load does not exceed expectations, the tasks to be executed are matched among the high-priority assembly tasks according to the current posture of the robot arm.

[0102] The path planning module plans the robot to move to the target location after the target location is determined by the task to be executed, and dynamically updates and marks it on the electronic map;

[0103] The motion control module predicts the robot's collision risk and adaptively adjusts the robot's motion state to at least make the end of the robot arm reach the target position;

[0104] A mode control module collects corresponding feedback data after automatically switching the assembly task, and switches the current assembly task switching mode if the feedback value generated by the feedback data does not exceed the expectation;

[0105] The safety control module generates a collision value from the recorded robot collision risk, sets a risk area in the assembly area when the collision value exceeds the expectation, and constrains the adjustment frequency of the risk area.

[0106] See also Figure 1-2 The present invention provides a dual-arm compound robot, which at least includes an upper arm and a lower arm, a motion joint providing multiple degrees of freedom, an end effector, a power unit for driving the joint and the arm movement; a sensor network; at least includes a position sensor, a torque sensor and a visual sensor.

[0107] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0108] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0109] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only some logical function divisions. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. The control system of the dual-arm compound robot is characterized by: include, The task scheduling module sorts the assembly tasks according to their priorities. If the assembly load does not exceed expectations, the task to be executed is matched among the high-priority assembly tasks according to the current posture of the robot arm. The current assembly task description data is collected, and the description data is used as input. The trained task scoring model is used to output the corresponding task score. After marking the assembly tasks with the task score, the assembly tasks are sorted from high to low and a task list is generated. The path planning module plans the target path for the robot to move to the target location after the target location is determined by the task to be executed, and dynamically updates and marks it on the electronic map; it obtains image data in the assembly area in real time, performs target detection and recognition on the acquired image data, obtains the type, volume and location data of the assembly parts in the assembly area, detects obstacles and the location data of assembly workers; constructs an electronic map covering the assembly area, and marks the acquired location data on the electronic map. The motion control module predicts the robot's collision risk and adaptively adjusts the robot's motion state to at least make the end of the robot arm reach the target position; A mode control module collects corresponding feedback data after automatically switching the assembly task, and switches the current assembly task switching mode if the feedback value generated by the feedback data does not exceed the expectation; The safety control module generates a collision value based on the recorded robot collision risk, sets a risk area in the assembly area when the collision value exceeds the expectation, and constrains the adjustment frequency of the risk area; the way to constrain the adjustment frequency of the risk area is as follows: ； Where: is the assembly cycle, is the distance threshold, The value is 2.718, is the time attenuation coefficient, ranging from 0 to 1. is the distance attenuation coefficient, ranging from 0.01 to 0.1; is the jth alarm node at time t, is the time interval between the ith and jth alarm nodes, is the number of alarm nodes, is the distance into the risk area at the jth alarm node; Collect feedback data when tasks are automatically switched, and construct feedback values ​​from the feedback data; if the feedback value does not exceed the preset feedback threshold, trigger the automatic mode switching mechanism to switch the automatic control mode of the current assembly task to the active control mode; receive the worker's voice commands through the microphone, and convert the worker's commands into control commands through semantic recognition and extraction; In the active control mode, the collision risk value of each sub-cycle is continuously collected, and the collision value is generated from the continuous collision risk values. If the collision value obtained exceeds the collision threshold, a risk area is set in the assembly area with the robot position as the center. The area of ​​the risk area is proportional to the ratio of the collision value exceeding the collision threshold, and the risk area is adjusted dynamically in real time; When the robot detects the approach of an assembler or an obstacle, it will automatically slow down; when the assembler enters the risk area, the robot's movement will automatically slow down or stop, the robotic arm will pause and send an alarm to the outside.

2. The control system of the dual-arm compound robot according to claim 1, characterized in that: The assembly task data in the current assembly cycle is collected, and the task density of the assembly tasks in the current assembly cycle is generated from the assembly task data set. If the task density does not exceed the density threshold, a new task addition permission is issued to the outside.

3. The control system of the dual-arm compound robot according to claim 2, characterized in that: After obtaining permission to add new tasks, dynamically adjust the task priority when new tasks are inserted; After obtaining several assembly tasks with higher task priorities in the task list, the robot arm's running trajectory is estimated when executing each corresponding assembly task, and the work task with the highest degree of matching with the robot arm's current state and running trajectory is selected as the task to be executed.

4. The control system of the dual-arm compound robot according to claim 3 is characterized in that: After selecting the task to be executed, the execution location of the task to be executed is used as the target location, and the Dijkstra algorithm is used to plan the shortest path from the starting location to the target location for the robot on the electronic map; The electronic map is updated in real time, the shortest path is adjusted according to the updated information, and the adjusted shortest path is used as the target path and marked on the electronic map.

5. The control system of the dual-arm compound robot according to claim 4, characterized in that: Obstacle data in the assembly area is collected, and the obstacle data is used as input to predict the robot's collision risk using the trained collision prediction model, and the corresponding collision risk value is output; When the collision risk value exceeds the preset risk threshold, the action of the robotic arm and the movement of the robot are stopped.

6. The control system of the dual-arm compound robot according to claim 5, characterized in that: The location of the assembly to be assembled or the expected arrival position is taken as the target position of the end of the robot arm. The inverse kinematics algorithm is used to calculate the angle and range of motion of each joint, and a PID controller is applied to each joint to adjust the motion trajectory of the robot arm in real time so that the end of the robot arm reaches the target position.

7. The control system of the dual-arm compound robot according to claim 6, characterized in that: The position and motion data of the assembly workers are collected and monitored in real time, and the real-time motion data of the robot arm and the current motion trajectory data are recorded and summarized as task status data; the assembly task with the highest degree of match with the current task status data is selected at the front of the task list as the task to be executed.

8. A dual-arm compound robot, controlled based on a control system according to any one of claims 1 to 7, characterized in that: At least including an upper arm and a lower arm, a motion joint providing multiple degrees of freedom, an end effector, and a power unit for driving the joint and the arm movement; The sensor network includes at least a position sensor, a torque sensor and a visual sensor.

Images