Unmanned aerial vehicle mechanical arm grasping pose planning method and system

Abstract

The present application relates to the technical field of mechanical arm path planning, in particular to a kind of unmanned plane mechanical arm grasping pose planning method and system.In the time when unmanned plane mechanical arm grasps object, the position parameter (including three-dimensional coordinates) and joint rotation angle time series data in each time node preset period are acquired;Due to the interference of strong wind, the attitude and motion fluctuation of mechanical arm deviate, analyze the change of joint rotation angle, random fluctuation characteristics, and the non-smooth fluctuation and numerical change of position parameter time series data, obtain the abnormality degree and deviation degree of strong wind interference, the degree of interference is quantified, the actual motion state is grasped;According to the two, adjust and fit the three-dimensional coordinate values of each time node, obtain the future extended trajectory, this method can respond to the change of strong wind, reduce trajectory error, provide accurate basis for path planning;Finally, based on the spatial position relationship of extended trajectory data points, the path is planned, the influence of strong wind is fully considered, the planned path is more accurate and reasonable, and the mechanical arm can be guided to optimally reach the target.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mechanical arm path planning, in particular to a UAV mechanical arm grasping pose planning method and system. BACKGROUND

[0002] With the rapid development of UAV technology, UAVs are increasingly widely used in many fields such as logistics transportation, disaster rescue, and environmental monitoring. In many complex task scenarios, UAVs need to be equipped with mechanical arms to achieve precise grasping of objects, such as power line inspection and rescue supplies in disaster rescue sites.

[0003] In the prior art, the least squares method is generally used to fit the motion trajectory of the position data and attitude data of the UAV mechanical arm, so as to extend and process it, for generating a grasping planning path. However, the UAV mechanical arm is easily disturbed by external strong wind factors during movement, and strong wind will have a significant impact on the flight attitude of the UAV and the movement of the mechanical arm, resulting in pose deviation. If the influence of this pose deviation is ignored and the trajectory is directly fitted, the error of the grasping planning path will be large, and the grasping accuracy will be greatly reduced, and even the grasping will fail. SUMMARY

[0004] In order to solve the technical problem that the UAV mechanical arm is easily disturbed by external strong wind factors during movement, and strong wind will have a significant impact on the flight attitude of the UAV and the movement of the mechanical arm, resulting in pose deviation, and if the influence of this pose deviation is ignored and the trajectory is directly fitted, the error of the grasping planning path will be large, and the grasping accuracy will be greatly reduced, and even the grasping will fail, the purpose of the present application is to provide a UAV mechanical arm grasping pose planning method and system, the technical scheme adopted is as follows:

[0005] A UAV mechanical arm grasping pose planning method, comprising:

[0006] In the process of the UAV mechanical arm grasping an object in a strong wind environment, the position parameter time series data and the rotation angle time series data of each joint corresponding to each time node in a preset time period are obtained, and the position parameter includes coordinate values in three spatial dimensions;

[0007] The change characteristics and random fluctuation characteristics of the rotation angle time series data of each joint corresponding to each time node are analyzed to determine the strong wind disturbance abnormality of the UAV mechanical arm at each time node;

[0008] Based on the non-smooth fluctuation characteristics and the change of the numerical value of the position parameter time series data corresponding to each time node, the strong wind disturbance deviation of the UAV mechanical arm in each spatial dimension at each time node is determined.

[0009] Adjust and fit the coordinate values of the three spatial dimensions at each time point based on the strong wind interference abnormality and the strong wind interference deviation, and obtain the extension trajectory of the UAV mechanical arm in the future period; based on the spatial position relationship of each data point on the extension trajectory, path planning is performed to obtain the grabbing planning path of the UAV mechanical arm.

[0010] Further, the method for obtaining the strong wind interference abnormality comprises:

[0011] In each joint rotation angle time series data within the preset period corresponding to each time node, the difference characteristics between the rotation angles at adjacent time points are analyzed to determine the sudden change rate of each joint at each time point;

[0012] In each time node corresponding to the preset period, the random fluctuation characteristics of the sudden change rate of the same joint at all time points are analyzed to obtain the strong wind interference factor corresponding to each joint at each time node;

[0013] The average value of the strong wind interference factors corresponding to all joints at each time node is taken as the strong wind interference abnormality of the UAV mechanical arm at each time node.

[0014] Further, the method for obtaining the sudden change rate comprises:

[0015] In the rotation angle time series data of each joint within the preset period corresponding to each time node, an optional time point is taken as a to-be-tested time point, the absolute value of the difference between the rotation angles at the to-be-tested time point and the adjacent previous time point is calculated as a change factor, the change factor is taken as the numerator, the sum of the rotation angle at the previous time point and a preset parameter is taken as the denominator, and the obtained ratio value is taken as the sudden change rate of each joint at the to-be-tested time point. The sudden change rate corresponding to the first time point is a preset value.

[0016] Further, the method for obtaining the strong wind interference factor comprises:

[0017] In each time node corresponding to the preset period, the sudden change rate sequence of each joint is obtained and subjected to STL time series decomposition to obtain the residual feature sequence corresponding to each joint.

[0018] The average value of the absolute values of the data values in the residual feature sequence corresponding to each joint is multiplied by the standard deviation of all data values in the residual feature sequence, and the normalized value of the obtained product is taken as the strong wind interference factor corresponding to each joint at each time node.

[0019] Further, the method for obtaining the strong wind interference deviation comprises:

[0020] In the preset time period corresponding to each time node, the coordinate value time series data in each spatial dimension is taken as the input of the moving smoothing algorithm to obtain the coordinate value smoothing sequence corresponding to each spatial dimension;

[0021] The difference features between the coordinate value time series data in each spatial dimension and the corresponding coordinate value smoothing sequence are compared to determine the coordinate non-smoothing deviation parameter of each time node in each spatial dimension;

[0022] In the preset time period corresponding to each time node, the first-order difference sequence of the coordinate value smoothing sequence in each spatial dimension is obtained, and the normalized value of the standard deviation of the absolute values of all data values in the first-order difference sequence is taken as the coordinate non-smoothing fluctuation parameter of each spatial dimension at each time node;

[0023] The sum of the coordinate non-smoothing deviation parameter and the coordinate non-smoothing fluctuation parameter of each time node in each spatial dimension is normalized to obtain the strong wind interference deviation degree of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time node.

[0024] Further, the coordinate non-smoothing deviation parameter acquisition method comprises:

[0025] In the coordinate value time series data and the corresponding coordinate value smoothing sequence corresponding to each spatial dimension, the absolute value of the difference value of the data values at the same time is calculated as a deviation factor, and the sum of the deviation factors at all times is normalized to obtain the coordinate non-smoothing deviation parameter of each time node in each spatial dimension.

[0026] Further, the extended trajectory acquisition method comprises:

[0027] After fusing the strong wind interference anomaly degree of the unmanned aerial vehicle mechanical arm at each time node and the strong wind interference deviation degree in each spatial dimension, the abnormal time nodes in each spatial dimension are screened from all time nodes;

[0028] In each spatial dimension, the coordinate values of the abnormal time nodes are corrected by using the polynomial interpolation method, so that the coordinate values of the abnormal time nodes are arranged in time sequence with the coordinate values of the non-abnormal time nodes to obtain the coordinate value sequence in each spatial dimension;

[0029] The coordinate values in the coordinate value sequence of each time node in all spatial dimensions are combined into three-dimensional coordinates, and the least squares method is used to curve fit the three-dimensional coordinates corresponding to all time nodes to obtain the historical motion trajectory of the unmanned aerial vehicle mechanical arm at the current time node;

[0030] The historical motion trajectory of the unmanned aerial vehicle mechanical arm at the current time node is processed by using the Bézier curve extension method to obtain the extended trajectory of the unmanned aerial vehicle mechanical arm in the future time period.

[0031] Further, the method for obtaining the abnormal time node comprises:

[0032] The strong wind interference abnormality of the unmanned aerial vehicle mechanical arm at all time points and the strong wind interference deviation in each spatial dimension are taken as inputs of the entropy weight method, so as to obtain the fusion weight corresponding to the strong wind interference abnormality and the strong wind interference deviation respectively, the strong wind interference abnormality and the strong wind interference deviation are weighted based on the fusion weight corresponding to the strong wind interference abnormality and the strong wind interference deviation, and the normalized value of the obtained weighted result is taken as the strong wind interference evaluation index of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time point.

[0033] The strong wind interference evaluation index of the unmanned aerial vehicle mechanical arm in each spatial dimension at all time nodes is taken as an input of the LOF anomaly detection algorithm, so as to obtain the abnormal time node.

[0034] Further, the method for obtaining the abnormal time node comprises:

[0035] The motion constraint is performed on the extended trajectory, so as to obtain all candidate coordinates corresponding to the constraint condition at each time in the future period;

[0036] The candidate coordinates at all times in the future period are projected onto a plane perpendicular to the plane in which the coordinate values of the three spatial dimensions of the current time node are located, all trajectory projection points are obtained and constitute a trajectory point set, and the Dijkstra algorithm is used to process the trajectory point set, so as to generate the grasping planning path of the unmanned aerial vehicle mechanical arm.

[0037] A grasping pose planning system of an unmanned aerial vehicle mechanical arm comprises a processor and a memory, and the memory stores at least one instruction, at least one program, a code set or an instruction set.

[0038] The present application has the following beneficial effects:

[0039] In the process of unmanned aerial vehicle mechanical arm grabbing objects, the position parameters (covering three spatial dimension coordinate values) and the joint rotation angle time series data in each time node corresponding to the preset period are obtained, which are used to provide information source for subsequent analysis of strong wind interference. When subjected to external strong wind interference, the attitude and movement of the unmanned aerial vehicle mechanical arm will usually produce certain fluctuations and deviations, so the change characteristics and random fluctuation characteristics of the rotation angle of each joint, and the non-smooth fluctuation characteristics and value changes of the position parameter time series data in each spatial dimension are analyzed, and the strong wind interference abnormality degree of the unmanned aerial vehicle mechanical arm in each time node and the strong wind interference deviation degree in each spatial dimension are obtained, which are used to quantify the interference degree of strong wind on the joint movement and movement position of the mechanical arm, and help to more accurately grasp the actual movement state of the mechanical arm in three-dimensional space. Further, the three spatial dimension coordinate values in each time node are adjusted and fitted according to the strong wind interference abnormality degree and the strong wind interference deviation degree, and the extension trajectory of the mechanical arm in the future period is obtained. This dynamic adjustment and fitting method can respond to the changes of strong wind interference, so that the extension trajectory is more in line with the actual movement trend of the mechanical arm in the strong wind environment, effectively reduces the trajectory error, and provides a more accurate basis for subsequent path planning. Finally, the path planning is carried out based on the spatial position relationship of each data point on the extension trajectory, and the grabbing planning path of the unmanned aerial vehicle mechanical arm is obtained. The path planning method fully considers the influence of strong wind interference on the movement of the mechanical arm, and the planned path is more accurate and reasonable, which can guide the mechanical arm to reach the target position in the optimal way. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, a brief introduction will be given to the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0041] Figure 1 A method flowchart of a unmanned aerial vehicle mechanical arm grabbing pose planning method provided by an embodiment of the present application;

[0042] Figure 2 A method flowchart of a strong wind interference abnormality degree acquisition method provided by an embodiment of the present application;

[0043] Figure 3 A method flowchart of a strong wind interference deviation degree acquisition method provided by an embodiment of the present application;

[0044] Figure 4 A method flowchart of an extension trajectory acquisition method provided by an embodiment of the present application;

[0045] Figure 5 Fig. 1 is a schematic diagram of a system structure of a UAV mechanical arm grasping pose planning system according to an embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to further illustrate the technical means and effects taken by the present application to achieve the predetermined purposes, the following describes in detail the specific embodiments, structures, features and effects of the UAV mechanical arm grasping pose planning method and system according to the present application, with reference to the accompanying drawings and preferred embodiments. Different "one embodiment" or "another embodiment" in the following description do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0048] The specific scheme of the UAV mechanical arm grasping pose planning method and system provided by the present application is described in detail below with reference to the accompanying drawings.

[0049] Please refer to Figure 1 which shows a method flowchart of a UAV mechanical arm grasping pose planning method according to an embodiment of the present application. The method comprises the following steps:

[0050] Step S1: During the process of the UAV mechanical arm grasping objects in a strong wind environment, the position parameter time series data and the rotation angle time series data of each joint in a preset time period corresponding to each time node are obtained. The position parameters include coordinate values in three spatial dimensions.

[0051] With the continuous development of artificial intelligence, robots and unmanned aerial vehicles, the application demand of three-dimensional pose estimation and grasping is increasing. For example, in the process of unmanned aerial vehicle power line inspection, the unmanned aerial vehicle is equipped with a UAV mechanical arm with multiple mechanical joints. The UAV mechanical arm needs to quickly capture moving target objects, which requires accurate path planning. However, the UAV is easily disturbed by strong wind in the actual working scene. Therefore, in the embodiment of the present application, the motion of the UAV mechanical arm in a strong wind environment is analyzed to eliminate the influence of external strong wind on the motion of the UAV mechanical arm, so as to achieve more accurate path planning.

[0052] Firstly, in a strong wind environment, a high-definition camera carried by a multi-rotor unmanned aerial vehicle collects the inspection video of the unmanned aerial vehicle mechanical arm in each time node corresponding to a preset period. The video collection process needs to ensure the angle adaptation between the high-definition camera and the inspection target to avoid video distortion caused by angle or position deviation. Then, multiple frames of inspection images are extracted from the inspection video of the unmanned aerial vehicle mechanical arm, and all the frames of inspection images are generated into an inspection image sequence. The inspection image sequence is used as a feature point detection algorithm, which can be a SIFT feature extraction algorithm or an ORB feature extraction algorithm. In this embodiment of the present application, the ORB feature extraction algorithm is used to extract the feature descriptors in the inspection image. Through multi-view geometric matching calculation, the three-dimensional spatial position of each feature point in the inspection image is obtained.

[0053] Further, the pose estimation is performed by using the PNP algorithm (Perspective-n-Point), and according to the known internal and external parameters of the high-definition camera, the three-dimensional spatial positions of all the feature points obtained by multi-view geometric matching, the pose parameter set of the unmanned aerial vehicle mechanical arm is estimated, and the pose parameter set of the unmanned aerial vehicle mechanical arm in each time node corresponding to a preset period is obtained, including position parameter time series data (coordinate values in X, Y and Z three spatial dimensions) and rotation angle time series data of each joint. The position parameter refers to the coordinate position of the center of mass of the unmanned aerial vehicle mechanical arm.

[0054] It should be noted that in the embodiment of the present application, the time node is selected to be the starting time of the history of 30 seconds, and each second is taken as a time node. The preset period corresponding to each time node is selected to be the starting time of the history of one minute. The collection frequency of various time series data is set to one second once. The ORB feature extraction algorithm, the PNP algorithm and the like are all known technologies, and the specific process will not be repeated here.

[0055] Step S2: analyze the change characteristics and random fluctuation characteristics of the rotation angle time series data of each joint corresponding to each time node, and determine the strong wind interference abnormality degree of the unmanned aerial vehicle mechanical arm under each time node.

[0056] When the unmanned aerial vehicle mechanical arm is disturbed by the sudden interference of external strong wind during the inspection process, the attitude of the unmanned aerial vehicle mechanical arm will appear sudden fluctuation, thereby affecting the stability of the actual pose parameters of the unmanned aerial vehicle mechanical arm. Therefore, in order to analyze the sudden fluctuation characteristics of the attitude of the unmanned aerial vehicle mechanical arm, the change characteristics and random fluctuation characteristics of the rotation angle time series data of each joint corresponding to each time node are analyzed in the embodiment of the present application, and the strong wind interference abnormality degree of the unmanned aerial vehicle mechanical arm under each time node is obtained, which is used to quantify the degree of abnormal change of the attitude of the unmanned aerial vehicle mechanical arm under the interference of external strong wind.

[0057] Preferably, the method for obtaining the strong wind interference abnormality degree in one embodiment of the present application comprises the following steps:

[0058] Referring to Figure 2 which shows a method flowchart of the method for obtaining the strong wind interference abnormality degree in one embodiment of the present application, the method comprises the following steps:

[0059] Step S201: In each joint rotation angle time series data within a preset time period corresponding to each time node, analyze the difference characteristics between the rotation angles at adjacent time points, and determine the sudden change rate of each joint at each time point.

[0060] Strong wind has randomness and suddenness, and its action on the robot arm joint will show as a sudden change in the rotation angle. Therefore, in the rotation angle time series data of each joint within a preset time period corresponding to each time node, the difference characteristics of the rotation angles at adjacent time points are analyzed. For ease of explanation and description, an optional time point is taken as the to-be-tested time point, and then the absolute value of the difference between the rotation angles at the to-be-tested time point and the adjacent previous time point is calculated as a change factor. The change factor can directly reflect the sudden change in the foregoing logic. The larger the change factor, the more dramatic the change in the rotation angle in a short period of time. The change factor is taken as the numerator, and the sum of the rotation angle at the previous time point and a preset parameter is taken as the denominator. The ratio eliminates the influence of the dimension, and the obtained ratio is taken as the sudden change rate of each joint at the to-be-tested time point. The larger the value, the more significant the change in the rotation angle at the to-be-tested time point compared with the previous time point, and the greater the sudden fluctuation degree, which means that the motion posture of the unmanned aerial vehicle robot arm at the to-be-tested time point is more unstable.

[0061] It should be noted that the preset parameter is used to prevent the denominator from being 0, and the value can be 0.2 degrees. The specific value can be adjusted according to the implementation scene, and is not limited herein. Since there is no previous adjacent time point for the first time point, the sudden change rate corresponding to the first time point is a preset value. The preset value in this embodiment of the present application is the average of the sudden change rates of all other time points. Thus, the sudden change rate of each joint at each time point within a preset time period corresponding to each time node can be obtained.

[0062] Step S202: Within a preset time period corresponding to each time node, analyze the random fluctuation characteristics of the sudden change rates of the same joint at all time points, and obtain the strong wind interference factor corresponding to each joint at each time node.

[0063] In order to measure the continuous characteristics of the abnormal changes of the unmanned aerial vehicle mechanical arm under the interference of external strong wind in the inspection process, the sudden change rate sequence under each joint in each time node corresponding to the preset period is obtained, the sudden change rate at each time is sorted in time sequence, and the residual feature sequence corresponding to each joint is obtained by STL time sequence decomposition. The residual feature sequence reflects the random fluctuation component in the sequence that cannot be explained by the trend and seasonality, and these random fluctuation components are often caused by external interference factors such as strong wind.

[0064] Therefore, the residual feature sequence is further analyzed: the mean of the absolute value of the data value in the residual feature sequence corresponding to each joint is calculated. The mean represents the average level of irregular changes of the unmanned aerial vehicle mechanical arm posture under the interference of external strong wind. The greater the value, the greater the interference degree. Then the standard deviation of all data values in the residual feature sequence is calculated. The standard deviation is used to reflect the dispersion degree. The greater the value, the lower the stability of the irregular changes of the unmanned aerial vehicle mechanical arm posture under the interference of external strong wind, which also indicates that the influence is more serious. Therefore, the product of the mean and the standard deviation is multiplied, and the normalized value of the product is taken as the strong wind interference factor corresponding to each joint at each time node. Based on the foregoing logic, the greater the strong wind interference factor, the more significant the influence of the posture of the joint on the unmanned aerial vehicle mechanical arm under the interference of external strong wind in the preset period corresponding to the time node. The normalization is a technology known to those skilled in the art. The selection of the normalization function can be linear normalization or standard normalization. The specific normalization method is not limited here.

[0065] Step S203: Synthesize the strong wind interference factors corresponding to all joints at each time node to determine the strong wind interference abnormality of the unmanned aerial vehicle mechanical arm at each time node.

[0066] In this sub-step, the mean of the strong wind interference factors corresponding to all joints at each time node can be calculated. The mean integrates the degree of interference of all joints on the unmanned aerial vehicle mechanical arm under the external strong wind at the same time node, so it can more comprehensively and accurately represent the degree of influence of the overall posture of the unmanned aerial vehicle mechanical arm at the time node. Therefore, the mean is taken as the strong wind interference abnormality of the unmanned aerial vehicle mechanical arm at each time node. The greater the strong wind interference abnormality, the more likely the unmanned aerial vehicle mechanical arm is to be subjected to sudden interference from external wind factors at the time node, which can affect the accuracy of the motion trajectory fitting of the unmanned aerial vehicle mechanical arm and is not conducive to the stability of the subsequent grasping of the target object by the unmanned aerial vehicle mechanical arm.

[0067] At this point, the strong wind interference abnormality of the unmanned aerial vehicle mechanical arm corresponding to each time node can be obtained.

[0068] Step S3: Based on the non-smooth fluctuation characteristics and the value changes of the position parameter time series data corresponding to each time node, the strong wind interference deviation degree of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time node is determined.

[0069] When the unmanned aerial vehicle mechanical arm is disturbed by external strong wind factors during the inspection process, the unmanned aerial vehicle mechanical arm movement is prone to non-smooth change characteristics, so that the unmanned aerial vehicle mechanical arm is prone to deviate from the smooth movement trajectory during the inspection process. Therefore, in order to more accurately eliminate the disturbance of the unmanned aerial vehicle mechanical arm by external strong wind factors during the movement process, thereby improving the accuracy of the movement trajectory fitting of the unmanned aerial vehicle mechanical arm, the non-smooth fluctuation characteristics of the position parameter time series data corresponding to each time node and the value changes thereof are analyzed in the embodiment of the application. The degree of deviation caused by the strong wind interference of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time node can be accurately quantified, that is, the strong wind interference deviation degree is obtained.

[0070] Preferably, the method for obtaining the strong wind interference deviation degree in one embodiment of the application comprises:

[0071] Please refer to Figure 3 which shows the method flowchart of the method for obtaining the strong wind interference deviation degree in one embodiment of the application. The method comprises the following steps:

[0072] Step S301: In a preset period corresponding to each time node, the coordinate value smooth sequence corresponding to the coordinate value time series data of each spatial dimension is obtained.

[0073] In the obtained coordinate value time series data of each spatial dimension, there will inevitably be noise caused by strong wind environment disturbance. By processing it through the moving smoothing algorithm, these random noises can be eliminated to obtain a more smooth coordinate sequence. The smooth sequence has a more regular form, which is convenient for subsequent difference analysis to more accurately capture abnormal changes caused by strong wind interference. Therefore, in the preset period corresponding to each time node, the coordinate value time series data in each spatial dimension is taken as the input of the moving smoothing algorithm, and the coordinate value smooth sequence corresponding to each spatial dimension is obtained.

[0074] It should be noted that the moving smoothing algorithm is a known technology, and the specific process is not described here.

[0075] Step S302: The difference characteristics between the coordinate value time series data of each spatial dimension and the corresponding coordinate value smooth sequence are compared, and the coordinate non-smooth deviation parameters of each time node in each spatial dimension are determined.

[0076] The strong wind has randomness and turbulent characteristics, and the action on the mechanical arm can cause short-term and irregular fluctuations in the position coordinates. The difference between the smooth sequence of coordinate values and the initial data can be used to measure the position deviation of the unmanned aerial vehicle mechanical arm in each spatial dimension under the influence of strong wind. Therefore, in the corresponding coordinate value time series data and the corresponding coordinate value smooth sequence in each spatial dimension, the absolute value of the difference between the data values at the same time is calculated as the deviation factor. The greater the deviation factor, the greater the degree of instantaneous change in the mechanical arm position under the interference of strong wind. The sum of the deviation factors at all times is normalized as the coordinate non-smooth deviation parameter of each time node in each spatial dimension. The coordinate non-smooth deviation parameter reflects the motion deviation characteristics of the unmanned aerial vehicle mechanical arm under the interference of external strong wind. The greater the value, the more likely the unmanned aerial vehicle mechanical arm deviates from the smooth motion trajectory under the interference of external strong wind. The normalization is a technical means familiar to those skilled in the art. The selection of the normalization function can be linear normalization or standard normalization, and the specific normalization method is not limited herein.

[0077] Step S303: In the preset time period corresponding to each time node, analyze the change of the coordinate value smooth sequence in each spatial dimension to determine the coordinate non-smooth fluctuation parameter of each spatial dimension at each time node.

[0078] Further, in order to accurately measure the non-smooth fluctuation change characteristics of the unmanned aerial vehicle mechanical arm under the influence of external strong wind, the first-order difference sequence of the coordinate value smooth sequence in each spatial dimension can be obtained in the preset time period corresponding to each time node. At this time, the absolute value of each data value in the first-order difference sequence represents the difference between adjacent two values in the coordinate value smooth sequence. The standard deviation of the absolute values of all data values in the first-order difference sequence is normalized as the coordinate non-smooth fluctuation parameter of each spatial dimension at each time node. The coordinate non-smooth fluctuation parameter reflects the non-smooth fluctuation component of the unmanned aerial vehicle mechanical arm under the influence of external strong wind. The greater the value, the greater the non-smooth fluctuation component of the unmanned aerial vehicle mechanical arm in the spatial dimension under the influence of external strong wind, which is more likely to cause the unmanned aerial vehicle mechanical arm to deviate from the smooth motion trajectory during the inspection process. The normalization is a technical means familiar to those skilled in the art. The selection of the normalization function can be linear normalization or standard normalization, and the specific normalization method is not limited herein.

[0079] Step S304: Fuse the coordinate non-smooth deviation parameter and the coordinate non-smooth fluctuation parameter of each time node in each spatial dimension to obtain the strong wind interference deviation degree of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time node.

[0080] Based on the analysis in steps S302 and S303, it can be known that the coordinate non-smooth deviation parameter and the coordinate non-smooth fluctuation parameter are positively correlated with the influence degree of the unmanned aerial vehicle mechanical arm in each space dimension under the influence of external strong wind at each time node. Therefore, in this embodiment, the sum of the coordinate non-smooth deviation parameter and the coordinate non-smooth fluctuation parameter in each space dimension at each time node is normalized, and the value after normalization is taken as the strong wind interference deviation degree of the unmanned aerial vehicle mechanical arm in each space dimension at each time node. The strong wind interference deviation degree reflects the significance of the motion deviation characteristics of the unmanned aerial vehicle mechanical arm in each space dimension under the influence of external strong wind. The greater the strong wind interference deviation degree, the more significant the motion deviation of the unmanned aerial vehicle mechanical arm in the corresponding space under the influence of external strong wind at the corresponding time node, and the more likely the unmanned aerial vehicle mechanical arm is to affect the accuracy of the motion trajectory fitting. The normalization is a technology known to those skilled in the art, and the normalization function can be linear normalization or standard normalization. The specific normalization method is not limited herein.

[0081] At this point, the strong wind interference deviation degree of the unmanned aerial vehicle mechanical arm in each space dimension at each time node can be obtained.

[0082] Step S4: adjusting and fitting the coordinate values of the three space dimensions at each time node based on the strong wind interference abnormality degree and the strong wind interference deviation degree, and obtaining the extended trajectory of the unmanned aerial vehicle mechanical arm in the future period; and performing path planning based on the spatial position relationship of each data point on the extended trajectory, so as to obtain the grasping planning path of the unmanned aerial vehicle mechanical arm.

[0083] The strong wind interference abnormality degree characterizes the influence degree of the overall posture of the unmanned aerial vehicle mechanical arm at the time node, and the strong wind interference deviation degree reflects the significance of the motion deviation characteristics of the unmanned aerial vehicle mechanical arm in each space dimension under the influence of external strong wind. In addition, the strong wind interference is dynamically changing, and the interference degree at different time nodes is also different. Therefore, the two indicators can be fused to adjust the coordinate values of the unmanned aerial vehicle mechanical arm in three-dimensional space at each time node, and then the trajectory fitting is performed. In this way, a more accurate motion trajectory of the unmanned aerial vehicle mechanical arm can be obtained, and the extended trajectory of the unmanned aerial vehicle mechanical arm in the future period can be obtained based on this, which can provide a more effective reference for the planning path of subsequent object grasping.

[0084] Preferably, in an embodiment of the present application, the method for obtaining the extended trajectory comprises:

[0085] Please refer to Figure 4 which shows a method flowchart of the method for obtaining the extended trajectory in an embodiment of the present application. The method comprises the following steps:

[0086] Step S401: After fusing the strong wind interference abnormality degree of the unmanned aerial vehicle mechanical arm at each time node and the strong wind interference deviation degree in each spatial dimension, the abnormal time nodes in each spatial dimension are screened out from all time nodes.

[0087] The strong wind interference abnormality degree and the strong wind interference deviation degree respectively reflect the influence of strong wind on the mechanical arm from different angles. The strong wind interference abnormality degree may focus on the abnormal influence degree of the overall environment on the mechanical arm posture, while the strong wind interference deviation degree pays more attention to the position deviation of the mechanical arm in each spatial dimension. The strong wind interference abnormality degree of the unmanned aerial vehicle mechanical arm at all time points and the strong wind interference deviation degree in each spatial dimension are taken as the input of the entropy weight method, so as to objectively weight the two indicators and obtain the fusion weight corresponding to the strong wind interference abnormality degree and the strong wind interference deviation degree respectively. The acquisition of the fusion weight can make the weight distribution in the subsequent fusion process more reasonable.

[0088] Then, the strong wind interference abnormality degree and the strong wind interference deviation degree are weighted based on the fusion weight corresponding to them, and the normalized value of the obtained weighted result is taken as the strong wind interference evaluation index of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time point. At this time, the strong wind interference evaluation index can more comprehensively evaluate the comprehensive influence of the unmanned aerial vehicle mechanical arm in each spatial dimension at each time point. The normalization is a technology known to those skilled in the art, and the selection of the normalization function can be linear normalization or standard normalization, and the specific normalization method is not limited here.

[0089] Finally, the strong wind interference evaluation index of the unmanned aerial vehicle mechanical arm in each spatial dimension at all time nodes is taken as the input of the LOF anomaly detection algorithm, so as to obtain the abnormal time nodes. The abnormal time nodes represent the significant outliers in all time nodes, which can be regarded as those time nodes most seriously affected by the strong wind interference. The position data in the abnormal time nodes is more seriously affected, so the possibility of inaccuracy is higher.

[0090] It should be noted that the entropy weight method and the LOF anomaly detection algorithm are both known technologies, and the specific process will not be described here.

[0091] Step S402: In each spatial dimension, the coordinate values in the abnormal time nodes are corrected to obtain the coordinate value sequence in each spatial dimension.

[0092] In each spatial dimension, the modified coordinate values at the abnormal time nodes are obtained by using a polynomial interpolation method and the coordinate values at the non-abnormal time nodes, and the coordinate values at the non-abnormal time nodes are arranged in time sequence, so as to obtain the coordinate value sequence in each spatial dimension. At this time, the coordinate value sequence eliminates the influence of strong wind disturbance on the robot arm, and thus more accurate results can be obtained based on the sequence for trajectory fitting and other processing.

[0093] It should be noted that the polynomial interpolation method is a known technology, and the specific process is not described herein.

[0094] Step S403: fitting and extending the coordinate value sequence in all spatial dimensions to obtain the extended trajectory of the UAV robot arm in the future period.

[0095] First, the coordinate values of each time node in the coordinate value sequence in all spatial dimensions are combined into three-dimensional coordinates, and then the least squares method is used to curve fit the three-dimensional coordinates corresponding to all time nodes, so as to obtain the historical motion trajectory of the UAV robot arm at the current time node.

[0096] The Bézier curve has good geometric properties and smoothness, and can flexibly describe various curve shapes. By using the Bézier curve extension method to extend the historical motion trajectory, the motion trajectory of the robot arm in the future period can be reasonably predicted according to the trend and characteristics of the historical motion trajectory, and the extension endpoint is the three-dimensional spatial coordinate of the target object (the grasped object). Therefore, the historical motion trajectory of the UAV robot arm at the current time node is processed by using the Bézier curve extension method, so as to obtain the extended trajectory of the UAV robot arm in the future period.

[0097] It should be noted that the least squares method and the Bézier curve extension method are known technologies, and the specific process is not described herein. In this embodiment of the present application, the length of the future period can be set to 5s, and the specific length can be adjusted according to the implementation scene, which is not limited herein.

[0098] The extended trajectory is obtained based on the prediction of the future motion state of the robot arm, and contains the position information of the robot arm at each time node in the future. Therefore, the path planning is performed based on the spatial position relationship of the data points on the extended trajectory, so as to fully utilize the prediction information, match the planned path with the actual motion trend of the robot arm, and improve the accuracy and effectiveness of the grasping planning path.

[0099] Preferably, in an embodiment of the present application, the method for obtaining the grasping planning path comprises:

[0100] First, the motion constraint is performed on the extension trajectory, and the constraint conditions include but are not limited to joint limit constraint and acceleration constraint. During the motion constraint process, the data points on the extension trajectory can be converted into a discrete point set, and thus all candidate coordinates corresponding to the constraint conditions at each time in the future time period can be obtained.

[0101] Then, the candidate coordinates at all times in the future time period are projected onto a plane perpendicular to the plane in which the coordinate values of the three spatial dimensions of the current time node are located, to obtain all trajectory projection points and form a trajectory point set. Finally, the Dijkstra algorithm is used to process the trajectory point set, to generate the grasping planning path (shortest path or optimal path) of the unmanned aerial vehicle mechanical arm.

[0102] It should be noted that the motion constraint algorithm and the Dijkstra algorithm are both known technologies, and the specific process will not be described here; the constraint conditions in the motion constraint can be determined by the hardware specifications and design structure of the unmanned aerial vehicle mechanical arm.

[0103] In summary, in the process of grasping an object by the unmanned aerial vehicle mechanical arm, the position parameters (covering the coordinate values of the three spatial dimensions) in the preset time period corresponding to each time node and the joint rotation angle time series data are obtained, which are used to provide an information source for subsequent analysis of strong wind interference. When subjected to external strong wind interference, the attitude and motion of the unmanned aerial vehicle mechanical arm will usually fluctuate and deviate to some extent. Therefore, the change characteristics and random fluctuation characteristics of the rotation angles of the joints, and the non-smooth fluctuation characteristics and value changes of the position parameter time series data in each spatial dimension are analyzed, to obtain the strong wind interference abnormality degree of the unmanned aerial vehicle mechanical arm corresponding to each time node and the strong wind interference deviation degree in each spatial dimension, which are used to quantify the interference degree of the strong wind on the joint motion and motion position of the mechanical arm, and help to more accurately grasp the actual motion state of the mechanical arm in the three-dimensional space. Further, the three spatial dimension coordinate values at each time node are adjusted and fitted according to the strong wind interference abnormality degree and the strong wind interference deviation degree, to obtain the extension trajectory of the mechanical arm in the future time period. This dynamic adjustment and fitting method can respond to the changes of the strong wind interference, so that the extension trajectory is more consistent with the actual motion trend of the mechanical arm in the strong wind environment, effectively reduces the trajectory error, and provides a more accurate basis for subsequent path planning. Finally, the path planning is performed based on the spatial position relationship of each data point on the extension trajectory, to obtain the grasping planning path of the unmanned aerial vehicle mechanical arm. This path planning method fully considers the influence of the strong wind interference on the motion of the mechanical arm, and the planned path is more accurate and reasonable, which can guide the mechanical arm to reach the target position in the optimal way.

[0104] The embodiment of the present application also provides a grasping pose planning system of an unmanned aerial vehicle mechanical arm, which is used for Figure 5Fig. 1 shows a schematic diagram of a system structure of a UAV mechanical arm grasping pose planning system according to an embodiment of the present application, which comprises a processor 500, a memory 501, a bus 502 and a communication interface 503, wherein the processor 500, the communication interface 503 and the memory 501 are connected through the bus 502; the memory 501 can contain a high-speed random access memory, the bus 502 can be an ISA bus, a PCI bus or an EISA bus, etc., the processor 500 can be an integrated circuit chip with signal processing capability; the memory 501 stores at least one instruction, at least one program, a code set or an instruction set, which are loaded and executed by the processor to implement steps of a UAV mechanical arm grasping pose planning method.

[0105] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0106] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

Claims

1. A method for planning the grasping pose of a drone robotic arm, characterized in that, The method includes: During the process of a drone robotic arm grasping an object in a strong wind environment, the timing data of the position parameters and the timing data of the rotation angle of each joint within a preset time period are obtained at each time node. The position parameters include coordinate values ​​in three spatial dimensions. Analyze the variation characteristics and random fluctuation characteristics of the rotation angle time series data of each joint at each time point to determine the strong wind interference anomaly degree of the UAV robotic arm at each time point. Based on the non-smooth fluctuation characteristics and numerical changes of the time series data of the location parameters corresponding to each time node, the deviation of the drone robotic arm from strong wind interference in each spatial dimension at each time node is determined. Based on the strong wind interference anomaly and strong wind interference deviation at each time point, the coordinate values ​​of the three spatial dimensions at each time point are adjusted and fitted, and the extension trajectory of the drone robotic arm in the future period is obtained; based on the spatial position relationship of each data point on the extension trajectory, path planning is performed to obtain the grasping planning path of the drone robotic arm. The method for obtaining the anomaly degree of strong wind interference includes: In the time series data of rotation angle of each joint within the preset time period corresponding to each time node, analyze the difference characteristics between rotation angles at adjacent time points, and determine the sudden change rate of each joint at each time point; Within a preset time period corresponding to each time node, the random fluctuation characteristics of the sudden change rate of the same joint at all times are analyzed to obtain the strong wind interference factor corresponding to each joint at each time node. The average value of the strong wind interference factor corresponding to all joints at each time point is taken as the strong wind interference anomaly degree of the drone robotic arm at each time point. The method for obtaining the strong wind interference deviation includes: Within a preset time period corresponding to each time node, the time series data of coordinate values ​​under each spatial dimension are used as input to the moving smoothing algorithm to obtain the smoothed sequence of coordinate values ​​corresponding to each spatial dimension. Compare the difference characteristics between the time series data of coordinate values ​​in each spatial dimension and the corresponding smoothed sequence of coordinate values ​​to determine the non-smoothing deviation parameters of coordinates at each time point in each spatial dimension; Within a preset time period corresponding to each time node, the first-order difference sequence of the smoothed coordinate value sequence under each spatial dimension is obtained, and the standard deviation of the absolute value of all data values ​​in the first-order difference sequence is normalized and used as the coordinate non-smooth fluctuation parameter for each spatial dimension under each time node. The normalized sum of the coordinate non-smooth deviation parameter and the coordinate non-smooth fluctuation parameter at each time point in each spatial dimension is used as the strong wind interference deviation of the drone robotic arm at each time point in each spatial dimension.

2. The method for planning the grasping pose of a UAV robotic arm according to claim 1, characterized in that, The method for obtaining the sudden change rate includes: In the rotation angle time series data of each joint within the preset time period corresponding to each time node, a time is randomly selected as the time to be measured. The absolute value of the difference between the rotation angle of the time to be measured and the adjacent previous time is calculated as the change factor. The change factor is used as the numerator, and the sum of the rotation angle of the previous time and the preset parameter is used as the denominator. The resulting ratio is used as the sudden change rate of each joint at the time to be measured. The sudden change rate corresponding to the first time is a preset value.

3. The method for planning the grasping pose of a UAV robotic arm according to claim 1, characterized in that, The method for obtaining the strong wind interference factor includes: Within the preset time period corresponding to each time node, the sudden change rate sequence under each joint is obtained and STL time series decomposition is performed to obtain the residual feature sequence corresponding to each joint. The mean of the absolute values ​​of the data values ​​in the residual feature sequence corresponding to each joint is multiplied by the standard deviation of all data values ​​in the residual feature sequence, and the normalized value of the resulting product is used as the strong wind interference factor for each joint at each time point.

4. The method for planning the grasping pose of a UAV robotic arm according to claim 1, characterized in that, The method for obtaining the coordinate non-smooth deviation parameter includes: In the time series data of coordinate values ​​corresponding to each spatial dimension and the corresponding smoothed sequence of coordinate values, the absolute value of the difference between the data values ​​at the same time is calculated as the deviation factor. The sum of the deviation factors at all times is normalized and used as the non-smoothing deviation parameter of the coordinates at each time node in each spatial dimension.

5. The method for planning the grasping pose of a UAV robotic arm according to claim 1, characterized in that, The method for obtaining the extended trajectory includes: After fusing the strong wind interference anomaly degree of the drone robotic arm at each time point and the strong wind interference deviation degree in each spatial dimension, the abnormal time points in each spatial dimension are selected from all time points. In each spatial dimension, the coordinate values ​​of abnormal time nodes are corrected using a polynomial interpolation method, and then arranged in chronological order with the coordinate values ​​of non-abnormal time nodes to obtain a sequence of coordinate values ​​in each spatial dimension. The coordinate values ​​of each time node in all spatial dimensions are combined into three-dimensional coordinates, and the least squares method is used to fit the three-dimensional coordinates corresponding to all time nodes to obtain the historical motion trajectory of the drone robotic arm at the current time node. The Bézier curve extension method is used to process the historical motion trajectory of the drone robotic arm at the current time point to obtain the extended trajectory of the drone robotic arm in the future time period.

6. The method for planning the grasping pose of a UAV robotic arm according to claim 5, characterized in that, The methods for obtaining the abnormal time nodes include: The strong wind interference anomaly degree of the drone robotic arm at all time points and the strong wind interference deviation degree in each spatial dimension are used as inputs of the entropy weight method to obtain the fusion weights corresponding to the strong wind interference anomaly degree and the strong wind interference deviation degree respectively. The fusion weights corresponding to the strong wind interference anomaly degree and the strong wind interference deviation degree are weighted, and the normalized value of the obtained weighted result is used as the strong wind interference evaluation index of the drone robotic arm in each spatial dimension at each time point. The strong wind interference evaluation index of the drone robotic arm in each spatial dimension at all time points is used as the input of the LOF anomaly detection algorithm to obtain the anomaly time points.

7. The method for planning the grasping pose of a UAV robotic arm according to claim 1, characterized in that, The method for obtaining the crawling planning path includes: Motion constraints are applied to the extended trajectory to obtain all candidate coordinates that meet the constraints at each time point in the future time period. Project the candidate coordinates at all times in the future time period onto a plane perpendicular to the coordinate values ​​of the three spatial dimensions of the current time node to obtain all trajectory projection points and form a trajectory point set. Use Dijkstra's algorithm to process the trajectory point set to generate the grasping planning path of the UAV robotic arm.

8. A robotic arm posture planning system for grasping unmanned aerial vehicles (UAVs), characterized in that, It includes a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, and when the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor, it implements the steps of the UAV robotic arm grasping pose planning method as described in any one of claims 1-7.

Images