Intelligent planning method for motion trail of multi-axis mechanical arm

Abstract

The invention discloses an intelligent planning method for a motion track of a multi-axis mechanical arm, and relates to the technical field of automatic control, and the intelligent planning method comprises the following steps: setting a task target and initial parameters of the mechanical arm, collecting environment data to generate an environment map, and constructing a dynamic environment model, the initial parameters of the mechanical arm comprise the initial position and posture, the joint angle, the speed and the acceleration; a mechanical arm track generation model is constructed, a mechanical arm preliminary track is generated based on the mechanical arm track generation model, the dynamic constraint of the mechanical arm preliminary track is checked, and the mechanical arm preliminary track is optimized; and predicting a future track of the obstacle according to the dynamic environment model, and combining the optimized preliminary track of the mechanical arm and the future track of the obstacle to generate a motion track of the mechanical arm. According to the method, through dynamic environment modeling, accurate pre-judgment of the motion trail of the dynamic obstacle is achieved, and the real-time adaptability of the mechanical arm in a complex scene is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automation control technology, in particular to a multi-axis robot motion trajectory intelligent planning method. BACKGROUND

[0002] With the rapid development of manufacturing and automation technology, multi-axis robots are increasingly widely used, from traditional industrial manufacturing to emerging service robot fields. Especially in high-precision, repetitive operations, multi-axis robots have shown irreplaceable advantages. Early robot control mainly relies on pre-programmed path planning, through a series of pre-set instructions to complete a specific task.

[0003] Most of the existing trajectory planning algorithms rely on pre-set static environment models, and often cannot update the motion path of the robot in real time to avoid obstacles, resulting in safety hazards and limiting the applicability of the robot in dynamic scenarios. In addition, the existing technology usually independently optimizes the dynamics constraints after generating the preliminary trajectory, without incorporating the task goal and robot initial parameters into a unified optimization framework. This fragmentation results in trajectories that may exceed the physical capabilities of the robot. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a multi-axis robot motion trajectory intelligent planning method to solve the problems of poor environmental adaptability and low trajectory feasibility.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a multi-axis robot motion trajectory intelligent planning method, which includes setting a task goal and robot initial parameters, collecting environment data to generate an environment map, and constructing a dynamic environment model. The task goal includes target point coordinates and a task time window, and the robot initial parameters include initial position and attitude, joint angle, speed and acceleration. A robot trajectory generation model is constructed, and a robot preliminary trajectory is generated based on the robot trajectory generation model. The dynamics constraints of the robot preliminary trajectory are checked, and the robot preliminary trajectory is optimized. The future trajectory of the obstacle is predicted according to the dynamic environment model, and the robot motion trajectory is generated in combination with the optimized robot preliminary trajectory and the future trajectory of the obstacle. The robot motion trajectory is executed, and the robot motion state is monitored in real time. The trajectory deviation is identified by direct comparison method, and the robot motion trajectory is adjusted in real time.

[0007] As a preferred solution of the multi-axis robot motion trajectory intelligent planning method of the present application, wherein: The setting task target and mechanical arm initial parameter, collecting environment data to generate environment map, constructing dynamic environment model, specific steps are as follows, The task target and mechanical arm initial parameter are stored as a task configuration file. The environment data includes point cloud data and image data. The point cloud data and image data are preprocessed, and a multi-modal registration method is used for fusion to generate a static environment map. The YOLOv5 model is used to identify dynamic obstacles, and the LSTM model is used to predict the motion state of the dynamic obstacles to generate dynamic obstacle state information. The static environment map and dynamic obstacle state information are integrated into a unified framework through a spatio-temporal alignment method to construct a dynamic environment model.

[0008] As a preferred scheme of the multi-axis mechanical arm motion trajectory intelligent planning method of the application, wherein: the mechanical arm trajectory generation model is constructed, and the mechanical arm preliminary trajectory is generated based on the mechanical arm trajectory generation model, and the specific steps are as follows, The input layer is defined based on the task configuration file, the feature extraction layer is defined based on the full connection layer and the ReLU activation function, and the trajectory planning layer is defined by using the RRT trajectory planning algorithm. The trajectory optimization layer is defined by using the gradient descent algorithm and the loss function. The mechanical arm trajectory generation model is constructed according to the input layer, the feature extraction layer, the trajectory planning layer and the trajectory optimization layer. The task configuration file is input into the mechanical arm trajectory generation model to generate the mechanical arm preliminary trajectory.

[0009] As a preferred scheme of the multi-axis mechanical arm motion trajectory intelligent planning method of the application, wherein: The dynamics constraint of the mechanical arm preliminary trajectory is checked, and the mechanical arm preliminary trajectory is optimized, and the specific steps are as follows, The dynamics constraint includes joint torque limit, joint speed limit and acceleration limit. The torque value, speed value and acceleration value of each joint of the mechanical arm at each time point on the mechanical arm preliminary trajectory are compared with the dynamics constraint. When the limit range is exceeded, the mechanical arm preliminary trajectory does not meet the dynamics constraint. The joint angle is optimized by using the linear interpolation algorithm, and the speed and acceleration of the joint are optimized by using the filtering algorithm.

[0010] As a preferred scheme of the multi-axis mechanical arm motion trajectory intelligent planning method of the application, wherein: The future trajectory of the obstacle is predicted according to the dynamic environment model, and the specific steps are as follows, The dynamic obstacle state information is input into the dynamic environment model, and the dynamic environment model combines a Kalman filter prediction algorithm to predict the future trajectory of the obstacle. The mean square error is used as an evaluation index of the future trajectory prediction result of the obstacle, and the dynamic environment model is updated according to the evaluation index.

[0011] As a preferred scheme of the multi-axis mechanical arm motion trajectory intelligent planning method, wherein: The mechanical arm motion trajectory is generated by combining the optimized preliminary trajectory of the mechanical arm and the future trajectory of the obstacle, and the specific steps are as follows, Based on the optimized preliminary trajectory of the mechanical arm, the distance between the mechanical arm and the obstacle is calculated by using the distance collision detection method, and collision risk analysis is performed. According to the collision risk analysis result, the preliminary trajectory of the mechanical arm is planned and smoothed to generate the mechanical arm motion trajectory.

[0012] As a preferred scheme of the multi-axis mechanical arm motion trajectory intelligent planning method, wherein: The mechanical arm motion trajectory is executed, and the motion state of the mechanical arm is monitored in real time, and the trajectory deviation is identified by a direct comparison method, and the specific steps are as follows, The mechanical arm motion trajectory and the real-time monitoring data are synchronized to a unified time axis through a data synchronization mechanism. The trajectory deviation is calculated by selecting the mechanical arm parameters at the same time point, the parameter deviation threshold is set according to the task requirements, and the trajectory deviation is compared with the parameter deviation threshold.

[0013] As a preferred scheme of the multi-axis mechanical arm motion trajectory intelligent planning method, wherein: The real-time adjustment of the mechanical arm motion trajectory refers to that when the trajectory deviation exceeds the parameter deviation threshold, it is determined as an abnormal situation, and the abnormal point is located and analyzed by a trajectory deviation analysis method, and the mechanical arm motion trajectory is adjusted in real time according to the analysis result.

[0014] In a second aspect, the present application provides a computer device, comprising a memory and a processor, and the memory stores a computer program, wherein: the computer program is executed by the processor to realize any step of the multi-axis mechanical arm motion trajectory intelligent planning method according to the first aspect of the present application.

[0015] In a third aspect, the present application provides a computer readable storage medium, which stores a computer program, wherein: the computer program is executed by the processor to realize any step of the multi-axis mechanical arm motion trajectory intelligent planning method according to the first aspect of the present application.

[0016] The application has the beneficial effects that: through dynamic environment modeling, the precise prediction of the dynamic obstacle motion trajectory is realized, the real-time adaptability of the mechanical arm in a complex scene is significantly improved, and the path conflict risk caused by environmental mutation is reduced; and through embedding the dynamic constraint into the trajectory planning process, the generated trajectory has physical feasibility and task efficiency optimality. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Fig. 1 A flowchart of the multi-axis mechanical arm motion trajectory intelligent planning method.

[0019] Fig. 2 A flowchart of constructing a dynamic environment model.

[0020] Fig. 3 A flowchart of constructing a mechanical arm trajectory generation model.

[0021] Fig. 4 A flowchart of real-time monitoring and adjustment. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned purposes, features and advantages of the application more apparent and easy to understand, the specific embodiments of the application will be described in detail below with reference to the drawings of the specification.

[0023] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, but the application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the application, therefore the application is not limited to the specific embodiments disclosed below.

[0024] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the application. In this specification, "in one embodiment" appearing in different places does not mean the same embodiment, nor is it an independent or selective embodiment that excludes other embodiments.

[0025] REFERENCE Figs. 1-4 For one embodiment of the application, the embodiment provides a multi-axis mechanical arm motion trajectory intelligent planning method, including the following steps: S1, set task target and initial parameters of the robot arm, collect environment data to generate an environment map, and construct a dynamic environment model. The task target includes target point coordinates and a task time window. The initial parameters of the robot arm include initial position and posture, joint angle, speed, and acceleration.

[0026] The task target and the initial parameters of the robot arm are stored as a task configuration file. It should be noted that the task target and the initial parameters of the robot arm are stored as a configuration file in the form of a key-value pair in JSON format, and the configuration file is saved in ASCII text format.

[0027] The environment data includes point cloud data and image data. It should be noted that the point cloud data refers to a set of surface points of all objects in the environment space obtained by a laser radar, and each surface point is represented by three-dimensional coordinates. The image data refers to two-dimensional visual information of the environment captured by an RGB camera, including pixel-level color (RGB value), texture, and coordinate information.

[0028] The point cloud data and the image data are preprocessed and fused using a multi-modal registration method to generate a static environment map. It should be noted that the point cloud data is denoised by removing outliers using a statistical filtering algorithm, and the image data is denoised by smoothing pixel noise using a Gaussian filter. The image pixel coordinate information is back-projected into camera three-dimensional coordinates using the camera intrinsic parameters, and the camera three-dimensional coordinates are converted to the laser radar coordinate system using the camera rotation matrix and translation vector. The geometric features in the point cloud data are extracted using the PCA algorithm, and the semantic features in the image data are extracted using the SIFT algorithm. The initial coordinate correspondence between the point cloud data and the image data is established based on the rotation matrix and the translation vector, and the rotation matrix and the translation vector are iteratively optimized using the ICP algorithm to minimize the Euclidean distance error between the corresponding points. Moving targets are detected using the YOLOv5 model, corresponding point clouds are filtered, static point clouds are down-sampled by voxel, overlapping regions are merged, and image data semantic features are associated with corresponding voxels to generate a static environment map.

[0029] It should also be noted that the rotation matrix describes the rotation relationship between the laser radar coordinate system and the camera coordinate system, indicating the rotation angle of the laser radar coordinate system around each axis. The translation vector describes the displacement of the origin of the laser radar coordinate system to the origin of the camera coordinate system, indicating the positional offset of the laser radar relative to the camera.

[0030] The YOLOv5 model is used to identify dynamic obstacles, and the LSTM model is used to predict the motion state of the dynamic obstacles to generate dynamic obstacle state information. It should be noted that the image data is input into the YOLOv5 model, the batch size, learning rate and training rounds of the YOLOv5 model are set, and CIoU is used as the loss function. The backbone network extracts features, and the Neck network fuses multi-scale features. Finally, the Head network outputs dynamic obstacle state information, calculates the loss function and updates the accuracy of the YOLOv5 model through the Adam optimizer until the maximum training rounds are reached; load the YOLOv5 model, set the size pixel of the input image and the confidence threshold, and input the image data into the YOLOv5 model. In the model, dynamic obstacle state information including category labels, dynamic obstacle bounding box center coordinates, dynamic obstacle timestamps, and confidence scores is generated. Based on the dynamic obstacle state information output by the YOLOv5 model, dynamic obstacle state information of consecutive frames is collected at fixed time intervals to form time series segments. Time series data is constructed and input into the LSTM model to generate position predictions for the next two steps. The current frame position detected by the YOLOv5 model and the future position predicted by the LSTM model are fused through Kalman filtering to generate dynamic obstacle state information.

[0031] The static environment map and dynamic obstacle status information are integrated into a unified framework through spatiotemporal alignment to construct a dynamic environment model. It should be noted that the timestamps of the static environment map and the timestamps of the dynamic obstacle state information are aligned to generate a unified time axis; the center coordinates of the dynamic obstacle bounding box are converted into the global coordinates of the static environment map through the built-in calibration parameters of the sensor to define the spatiotemporal alignment layer; if the update time interval of the dynamic obstacle state information is inconsistent with the update time interval of the static environment map, the dynamic obstacle state information is downsampled or interpolated to adjust the time resolution; the center coordinates of the dynamic obstacle bounding box are aligned to the global coordinates of the static environment map through the coordinate system conversion alignment method, and the continuous coordinates are mapped to discrete grid units using the floor function to form the dynamic obstacle grid coordinates. If the dynamic obstacle If the center coordinates of the bounding box exceed the range of the static environment map, the grid boundaries of the static environment map are extended, and the grid coordinates of the dynamic obstacle are aligned with the grid coordinates of the static environment map to form a unified spatial resolution. The geometric features and speech features in the static environment map are mapped to a unified time axis, and the dynamic obstacle state information is inserted at the corresponding timestamp. The static environment map and the dynamic obstacle state information are merged to form a dynamic environment model that contains all obstacle information at the timestamp and global coordinates. Each time new dynamic obstacle state information arrives, only the dynamic obstacle state information corresponding to the timestamp is updated. The geometric features and semantic features in the static environment map are only refreshed when the static environment map is updated.

[0032] It should be noted that the dynamic environment model uses mean square error as the loss function, the Adam optimizer, sets the initial learning rate and the training round, inputs the static environment map and dynamic obstacle state information into the dynamic environment model for training until the mean square error reaches the minimum value, and generates the trained dynamic environment model.

[0033] S2, a robot arm trajectory generation model is constructed, a robot arm preliminary trajectory is generated based on the robot arm trajectory generation model, a dynamic constraint of the robot arm preliminary trajectory is checked, and the robot arm preliminary trajectory is optimized.

[0034] An input layer is defined based on the task configuration file, a feature extraction layer is defined based on a fully connected layer and a ReLU activation function, and a trajectory planning layer is defined using an RRT trajectory planning algorithm; It should be noted that the number of input layer nodes is determined according to the type and dimension of the target point coordinates and the initial parameters of the robot arm in the task configuration file; a plurality of fully connected layers are used to realize feature extraction, each fully connected layer contains a certain number of neurons, the neurons are connected through weights and biases, and a ReLU activation function is used between the fully connected layers to realize nonlinear characteristics; the RRT trajectory planning algorithm defines the node structure of the tree, each node contains information such as the joint angle and position of the robot arm, generates a new node in the configuration space through random sampling, and connects the new node with the existing nodes in the tree, realizes the logic of the RRT trajectory planning algorithm, generates an initial obstacle avoidance trajectory using the RRT trajectory planning algorithm, and the initial obstacle avoidance trajectory includes trajectory parameters (initial position, velocity and acceleration); the trajectory planning layer has a plurality of MLP layers, each MLP layer contains a certain number of neurons, and is connected through weights and biases.

[0035] A trajectory optimization layer is defined using a gradient descent algorithm and a loss function; It should be noted that the positioning loss function is defined according to the initial obstacle avoidance path and the target point coordinates; the gradient of the positioning loss function with respect to the trajectory parameters is calculated by back propagation, the initial learning rate is set to 0.01, and the decay strategy (decaying to 0.9 times the original value of the trajectory parameters every one hundred iterations) is configured to update the trajectory parameters.

[0036] A robot arm trajectory generation model is constructed according to the input layer, the feature extraction layer, the trajectory planning layer and the trajectory optimization layer; It should be noted that the task configuration file is divided into a training set, a validation set and a test set; all parameters of the robot arm trajectory generation model (weights and biases of the fully connected layer and trajectory parameters) are randomly initialized, the training rounds, batch size and learning rate of the robot arm trajectory generation model are set; the training set data is divided according to the set batch size and input into the robot arm trajectory generation model, and the robot arm preliminary trajectory is output by the trajectory optimization layer through forward propagation; the robot arm dynamics simulation platform Gazebo is used to simulate the robot arm motion, generate the robot arm simulation trajectory, calculate the error between the robot arm preliminary trajectory and the robot arm simulation trajectory using the positioning loss function, calculate the gradient of the loss to the parameters of each layer of the robot arm trajectory generation model through the back propagation algorithm, start from the output layer, calculate the gradient of each layer in turn, until the gradient of each layer is calculated, and the parameters of the robot arm trajectory generation model are updated using the Adam optimizer according to the calculated gradient, the Adam optimizer adjusts the learning rate according to the gradient information of each parameter, and calculates the first moment correction and second moment correction of the gradient to update the weights and biases of the robot arm trajectory generation model; after reaching the set training rounds, the validation set is input into the robot arm trajectory generation model, and the loss value and accuracy of the robot arm trajectory generation model on the validation set are calculated; if the performance of the robot arm trajectory generation model on the validation set no longer improves, the learning rate and batch size of the robot arm trajectory generation model need to be adjusted; the test set is input into the robot arm trajectory generation model, and whether the robot arm trajectory generation model reaches the expected effect is judged according to the trajectory accuracy, the expression is, ; wherein, represents the accuracy of the robot arm preliminary trajectory; represents the total number of trajectory points of the robot arm preliminary trajectory; represents the index variable of the trajectory point of the robot arm preliminary trajectory, and the value range is 1- ; represents the coordinate of the trajectory point of the robot arm trajectory generation model; represents the coordinate of the target point of the task target; represents the error tolerance parameter, and the value range is 0.05-0.2; when , it represents that the robot arm trajectory generation model meets the standard; when , it represents that the robot arm trajectory generation model does not meet the standard, and the number of iterations needs to be increased or the learning rate needs to be reduced.

[0037] The task configuration file is input into the robot arm trajectory generation model to generate a robot arm preliminary trajectory; The dynamics constraints include joint torque limit, joint speed limit and acceleration limit; ​It should be noted that the joint torque limit refers to the maximum torque value that each joint can safely output, exceeding which can cause motor overload or mechanical structure damage; the joint speed limit refers to the maximum angular velocity allowed for joint movement, avoiding mechanical arm-induced vibration or loss of control; the acceleration limit refers to the maximum allowed joint angular acceleration, avoiding mechanical impact or control delay caused by sudden acceleration or deceleration.

[0038] The joint torque value, joint speed value, and joint acceleration value of each joint of the mechanical arm at each time point on the preliminary trajectory of the mechanical arm are compared with the maximum allowed values; It should be noted that according to the mechanical arm product specification, the maximum allowed torque value, the maximum allowed speed value, and the maximum allowed acceleration value of the mechanical arm are determined. For each joint at each time point on the preliminary trajectory of the mechanical arm, the current joint torque value is calculated using the dynamics equation, the current joint speed value and the current joint acceleration value are calculated using the kinematics differential equation, and the calculated joint torque value, joint speed value, and joint acceleration value are compared with the maximum allowed torque value, maximum allowed speed value, and maximum allowed acceleration value of the corresponding joint one by one. If any of the joint torque value, joint speed value, and joint acceleration value exceeds the maximum allowed value limit, the joint is marked as violating the dynamics constraint.

[0039] When the limit range is exceeded, the preliminary trajectory of the mechanical arm does not satisfy the dynamics constraint; A linear interpolation algorithm is used to optimize the joint angle, and a filtering algorithm is used to optimize the joint speed and acceleration; It should be noted that for each over-limit joint, the adjacent non-over-limit time points in the preliminary trajectory of the mechanical arm are selected as interpolation reference points, and the corrected joint torque value at the over-limit time point is calculated using a linear interpolation algorithm, expressed as, ; wherein, represents the optimized joint torque value; represents the joint torque value at the start of interpolation, represents the joint torque value at the end of interpolation; represents the time at the current time point that needs to be optimized; represents the time at the start of interpolation; represents the time at the end of interpolation; A second-order low-pass filter is used to smooth the joint speed, expressed as, ; wherein, represents the optimized speed value, represents the speed value at the current time point ; represents the speed value at the previous time point of the current time point ; represents the current time point; represents the previous time point of the current time point; represents the joint speed smoothing coefficient, the value range is is a parameter for controlling the filtering strength, the smaller the value is, the smoother the filtered speed value is; The Savitzky filter is used to optimize the joint acceleration, and the expression is ; wherein, represents the optimized acceleration value; is the filter window size, representing the current time point participate in the calculation of the acceleration value within the range of time steps before and after the current time point; is the half-width of the filter window, representing the range of time steps before and after the current time point; represents the relative time index in the filter window, which is used to traverse the acceleration values within the range of time steps before and after the current time point; to ; represents the fitting coefficient, which is used to perform weighted average on the acceleration values in the filter window; represents the acceleration value at the current time point ; represents the current time point.

[0040] S3, predicting the future trajectory of the obstacle according to the dynamic environment model, combining the optimized preliminary trajectory of the robot arm and the future trajectory of the obstacle to generate the robot arm motion trajectory.

[0041] The dynamic obstacle state information is input into the dynamic environment model, and the dynamic environment model combines the Kalman filter prediction algorithm to predict the future trajectory of the obstacle; ​​It should be pointed out that each dynamic obstacle state information is organized into a dynamic obstacle state vector containing position, velocity and acceleration, and a dynamic obstacle state transition matrix is established according to the motion characteristics (uniform motion or uniform acceleration motion) of the dynamic obstacle; the Kalman filter prediction algorithm is applied, the dynamic obstacle state prediction value at the next time is calculated through the dynamic obstacle state transition matrix, the dynamic obstacle state value of the actual motion of the dynamic obstacle at the current time is obtained through the sensor, the Kalman gain is calculated, the dynamic obstacle state vector is updated according to the Kalman gain, the updated dynamic obstacle state vector is used to recursively calculate the dynamic obstacle state prediction vector at multiple future time steps through the dynamic obstacle state transition matrix, the dynamic obstacle state prediction vector is converted into a specific trajectory point sequence through the time step method, and each trajectory point contains position, velocity and acceleration information.

[0042] The mean square error is used as an evaluation index of the future trajectory prediction result of the obstacle, and the dynamic environment model is updated according to the evaluation index; It should be pointed out that the mean square error is calculated according to the dynamic obstacle state prediction vector and the dynamic obstacle state value observed by the sensor, and the expression is, ; Among them, is the mean square error, which represents the deviation of the dynamic obstacle state prediction vector and the dynamic obstacle state value observed by the sensor; represents the total number of time steps participating in error calculation; represents the time step index, and the value range is 1- ; represents the prediction error vector of the th time step, , represents the dynamic obstacle state prediction vector of the th time step, represents the dynamic obstacle state value observed by the sensor at the th time step; The error threshold is set to 0.52, when , the diagonal elements of the dynamic obstacle state transition matrix are increased, the updated dynamic environment model is used to generate a new dynamic obstacle prediction trajectory, and is recalculated, , the latest future trajectory of the obstacle is generated.

[0043] Based on the optimized preliminary trajectory of the mechanical arm, the distance between the mechanical arm and the obstacle is calculated by using the distance collision detection method, and the collision risk analysis is carried out; It should be noted that the trajectory points of the preliminary trajectory of the manipulator are divided into discrete point sets according to the same time step, and the minimum Euclidean distance between each trajectory point and all obstacles is calculated. The minimum safe distance between the end of the manipulator and the obstacle surface is calculated based on the minimum Euclidean distance; the static safety threshold is set to 0.2 meters, and when the minimum safe distance When the minimum safe distance is 0.2 meters, the risk level is low risk. When the distance is 0.2 meters, the risk level is high.

[0044] Plan the preliminary trajectory of the robot arm based on the collision risk analysis results, perform smoothing, and generate the robot arm motion trajectory; It should be noted that if the risk level is low, the original trajectory points of the robot arm's preliminary trajectory remain unchanged; if the risk level is high, the reverse vector from the trajectory point of the minimum safe distance to the obstacle is calculated, and the position of the trajectory point is adjusted to keep the robot arm away from the obstacle; the adjusted preliminary trajectory of the robot arm is divided into several segments, each segment contains continuous trajectory points, and the Bezier curve fitting smoothing algorithm is applied to each segment to smooth the preliminary trajectory of the robot arm and generate the robot arm motion trajectory.

[0045] S4. Execute the robot arm motion trajectory and monitor the robot arm motion status in real time. Identify the trajectory deviation through direct comparison method and adjust the robot arm motion trajectory in real time.

[0046] The robot arm motion trajectory and real-time monitoring data are synchronized to a unified time axis through the data synchronization mechanism; It should be noted that the timestamps of all devices are synchronized through the network time protocol; the robot arm motion trajectory and real-time monitoring data are generated into corresponding continuous time series according to the linear interpolation method; for each monitoring data point in the real-time monitoring data time series, the corresponding time point is found on the robot arm motion trajectory time series, and a corresponding relationship is established for real-time collision detection and environmental interaction analysis. When there is nonlinear distortion in the time series, the dynamic time warping algorithm is used to calculate the optimal time alignment path; the interpolated robot arm motion trajectory and real-time monitoring data are merged according to a unified time axis through the timestamp matching method to form a comprehensive data set containing information such as position, speed, and obstacle status.

[0047] Select the robot arm parameters at the same time point to calculate the trajectory deviation, set the parameter deviation threshold according to the task requirements, and compare the trajectory deviation with the parameter deviation threshold; It should be noted that the trajectory deviation includes a spatial distance deviation, a relative speed deviation and an angle deviation, and the parameter deviation threshold includes a distance deviation threshold, a speed deviation threshold and an angle deviation threshold; the mechanical arm movement trajectory parameters including position coordinates, speed components and joint angles and the obstacle state parameters including obstacle positions, speed components and size parameters are extracted from the comprehensive data set; the trajectory deviation between the mechanical arm end effector and the obstacle is calculated; the parameter deviation threshold is set according to the task requirements, the calculated trajectory deviation is compared with the parameter deviation threshold, if each item of the trajectory deviation is less than the parameter deviation threshold, it is determined that the mechanical arm movement trajectory is safe, if any item of the trajectory deviation is out of limit, it is determined that the mechanical arm movement trajectory has risks and needs to be adjusted.

[0048] The real-time adjustment of the mechanical arm movement trajectory means that when the trajectory deviation exceeds the parameter deviation threshold, it is determined as an abnormal situation, the abnormal point is positioned and analyzed through the trajectory deviation analysis method, and the mechanical arm movement trajectory is adjusted in real time according to the analysis result; It should be noted that the time stamp of the abnormal situation is recorded, which is accurate to the millisecond level time unit, the spatial position of the mechanical arm end effector when the abnormal situation occurs is determined according to the position coordinates in the mechanical arm movement trajectory, and the joint angle and angular velocity state parameters when the abnormal situation occurs are extracted; whether the trajectory deviation is caused by the obstacle approaching is analyzed, and whether the joint torque, speed and acceleration when the abnormal situation occurs exceed the physical limit of the mechanical arm is verified; the adjustment mode is selected according to the abnormal reason, if it is because the obstacle approaches, the mechanical arm movement trajectory needs to be re-planned to bypass the obstacle, if it is because the physical limit of the mechanical arm is exceeded, the dynamics constraint of the mechanical arm needs to be re-adjusted; the adjusted mechanical arm movement trajectory is smoothed by using a Bezier curve.

[0049] The embodiment also provides a computer device suitable for the multi-axis mechanical arm movement trajectory intelligent planning method, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the multi-axis mechanical arm movement trajectory intelligent planning method proposed in the above embodiment.

[0050] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is configured to perform wired or wireless communication with an external terminal. The wireless communication can be achieved by WIFI, an operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0051] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the method for intelligently planning a motion trajectory of a multi-axis robot arm according to the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0052] To sum up, the application achieves accurate prediction of the motion trajectory of a dynamic obstacle by dynamic environment modeling, significantly improves the real-time adaptability of the robot arm in a complex scene, and reduces the risk of path conflict caused by environmental mutation; and by embedding the dynamic constraint into the trajectory planning process, the generated trajectory has both physical feasibility and optimal task efficiency.

[0053] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. An intelligent planning method for multi-axis robotic arm motion trajectory, characterized by: include, Set the mission objectives and initial parameters of the manipulator, collect environmental data to generate an environmental map, and build a dynamic environment model. The mission objectives include target point coordinates and mission time windows, and the initial parameters of the manipulator include initial position and posture, joint angles, speed, and acceleration. Construct a robot arm trajectory generation model, generate a preliminary robot arm trajectory based on the robot arm trajectory generation model, check the dynamic constraints of the preliminary robot arm trajectory, and optimize the preliminary robot arm trajectory; The future trajectory of the obstacle is predicted based on the dynamic environment model, and the robot arm motion trajectory is generated by combining the optimized preliminary trajectory of the robot arm and the future trajectory of the obstacle; Execute the robot arm motion trajectory and monitor the robot arm motion status in real time. Identify trajectory deviations through direct comparison method and adjust the robot arm motion trajectory in real time.

2. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 1, wherein: The specific steps of setting the mission objectives and initial parameters of the robotic arm, collecting environmental data to generate an environmental map, and building a dynamic environmental model are as follows: Store the mission objectives and initial parameters of the robotic arm as a mission configuration file; Environmental data includes point cloud data and image data; Preprocess the point cloud data and image data and fuse them using multimodal registration methods to generate a static environment map; Use the YOLOv5 model to identify dynamic obstacles, and use the LSTM model to predict the motion state of dynamic obstacles and generate dynamic obstacle status information; The static environment map and dynamic obstacle status information are integrated into a unified framework through the spatiotemporal alignment method to construct a dynamic environment model.

3. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 2, wherein: The robot arm trajectory generation model is constructed, and the preliminary trajectory of the robot arm is generated based on the robot arm trajectory generation model. The specific steps are as follows: Define the input layer based on the task configuration file, define the feature extraction layer based on the fully connected layer and ReLU activation function, and define the trajectory planning layer using the RRT trajectory planning algorithm; Define the trajectory optimization layer using the gradient descent algorithm and loss function; Construct a robot trajectory generation model based on the input layer, feature extraction layer, trajectory planning layer, and trajectory optimization layer; The task configuration file is input into the robot trajectory generation model to generate a preliminary trajectory for the robot.

4. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 3, wherein: The steps of checking the dynamic constraints of the preliminary trajectory of the manipulator and optimizing the preliminary trajectory of the manipulator are as follows: Dynamic constraints include joint torque limits, joint velocity limits, and acceleration limits; Compare the torque value, velocity value and acceleration value of each joint of the manipulator at each time point on the preliminary trajectory of the manipulator with the dynamic constraints; When the limit range is exceeded, the initial trajectory of the robot arm does not satisfy the dynamic constraints; The linear interpolation algorithm is used to optimize the joint angles, and the filtering algorithm is used to optimize the joint velocity and acceleration.

5. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 4, wherein: The specific steps of predicting the future trajectory of obstacles based on the dynamic environment model are as follows: The dynamic obstacle state information is input into the dynamic environment model, and the dynamic environment model is combined with the Kalman filter prediction algorithm to predict the future trajectory of the obstacle; The mean square error is used as the evaluation index of the obstacle's future trajectory prediction results, and the dynamic environment model is updated according to the evaluation index.

6. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 5, wherein: The specific steps of combining the optimized preliminary trajectory of the robot arm and the future trajectory of the obstacle to generate the robot arm motion trajectory are as follows: Based on the optimized preliminary trajectory of the robotic arm, the distance between the robotic arm and the obstacle is calculated using the distance collision detection method, and collision risk analysis is performed; The preliminary trajectory of the robot arm is planned according to the collision risk analysis results, and smoothing is performed to generate the robot arm motion trajectory.

7. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 6, wherein: The execution of the robot arm motion trajectory and the real-time monitoring of the robot arm motion state are performed, and the trajectory deviation is identified by direct comparison method. The specific steps are as follows: The robot arm motion trajectory and real-time monitoring data are synchronized to a unified time axis through the data synchronization mechanism; The robot arm parameters at the same time point are selected to calculate the trajectory deviation, the parameter deviation threshold is set according to the task requirements, and the trajectory deviation is compared with the parameter deviation threshold.

8. The method for intelligently planning motion trajectories of a multi-axis robotic arm according to claim 1, wherein: The real-time adjustment of the robot arm motion trajectory means that when the trajectory deviation exceeds the parameter deviation threshold, it is determined to be an abnormal situation, and the abnormal point is located and analyzed through the trajectory deviation analysis method, and the robot arm motion trajectory is adjusted in real time according to the analysis results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-axis robotic arm motion trajectory intelligent planning method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-axis robotic arm motion trajectory intelligent planning method according to any one of claims 1 to 8 are implemented.

Images