Method and device for determining the position of the end of a flexible arm, and flexible arm mechanical equipment

Abstract

The present application discloses a method, device and flexible arm mechanical equipment for determining the end position of a flexible arm. The method includes: obtaining the joint angle of the target joint of the flexible arm and the three-axis speed of the end of the flexible arm at the current moment; inputting the joint angle into a neural network model to obtain the predicted deformation of the end of the flexible arm; performing coordinate transformation on the predicted deformation of the end to obtain the predicted deformation of the device; establishing an observation matrix based on the predicted deformation of the device to obtain the theoretical end position; establishing a state transition model based on the three-axis speed to obtain the predicted end position; performing Kalman filtering on the theoretical end position and the predicted end position to determine the final end position of the flexible arm. The present application improves the prediction accuracy of the end position of the flexible arm mechanical equipment, thereby improving the efficiency of the automatic operation of the flexible arm mechanical equipment.

Description

Technical Field

[0001] The present application relates to the technical field of mechanical equipment, and in particular to a method and device for determining the end position of a flexible arm, and a flexible arm mechanical equipment. Background Art

[0002] For nonlinear systems such as the flexible booms of construction machinery such as pump trucks and concrete placing booms, which are highly redundant, multivariable, and have complex dynamic coupling characteristics, the controllers currently used cannot control the entire nonlinear flexible arm as a whole during the control process. Moreover, as the boom's movement speed increases, the accuracy of the boom end trajectory tracking cannot meet actual engineering requirements. The existing technology simplifies the boom into a rigid body, and the corresponding joint angles can be obtained through the rigid inverse kinematics of the boom. However, the actual boom undergoes large elastic deformation under the influence of gravity, resulting in the boom end being unable to accurately reach the target distribution point. Therefore, in the existing technology, the prediction accuracy of the end position of the flexible arm mechanical equipment is low. Summary of the Invention

[0003] The purpose of the embodiments of the present application is to provide a method, an apparatus, and a flexible arm mechanical device for determining the end position of a flexible arm, so as to solve the problem of low accuracy in predicting the end position of a flexible arm mechanical device in the prior art.

[0004] To achieve the above objectives, the present application provides, in a first aspect, a method for determining the position of an end of a flexible arm, which is applied to a flexible arm mechanical device, and comprises:

[0005] Get the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment;

[0006] The joint angles are input into the neural network model to obtain the predicted deformation of the end of the flexible arm;

[0007] Perform coordinate transformation on the predicted deformation of the terminal to obtain the predicted deformation of the equipment;

[0008] Establish an observation matrix based on the predicted deformation of the equipment to obtain the theoretical end position;

[0009] Establish a state transfer model based on the three-axis speed to obtain the predicted end position;

[0010] The theoretical end position and the predicted end position are Kalman filtered to determine the final end position of the flexible arm.

[0011] In the embodiment of the present application, obtaining the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment includes:

[0012] Acquiring a joint angle of a target joint by means of an inclination sensor or a displacement sensor provided at the target joint of the flexible arm;

[0013] The three-axis velocity of the end of the flexible arm is obtained by an inertial navigation sensor arranged at the end of the flexible arm.

[0014] In the embodiment of the present application, the joint angle is input into the neural network model to obtain the predicted deformation of the end of the flexible arm, including:

[0015] Input joint angles into multiple neural network models with different structures;

[0016] Obtain multiple terminal predicted deformation components output by multiple neural network models;

[0017] The middle value of the output multiple terminal predicted deformation components is selected as the reference value of the normal distribution function;

[0018] The weight corresponding to each terminal predicted deformation component is obtained according to the reference value and the normal distribution function;

[0019] Each terminal predicted deformation component and the corresponding weight are weighted squared to obtain the terminal predicted deformation amount.

[0020] In the embodiment of the present application, the neural network model is a full neural network model, and the full neural network model satisfies the following formula:

[0021]

[0022] in, is the activation value of neuron i in layer l; is the neuron in layer l-1; For neurons The corresponding weight; is the offset.

[0023] In the embodiment of the present application, performing coordinate transformation on the terminal predicted deformation to obtain the device predicted deformation includes:

[0024] Get the angle of the rotation joint of the rotating platform at the current moment;

[0025] The predicted deformation of the terminal is converted into coordinates according to the angle to obtain the predicted deformation of the device.

[0026] In the embodiment of the present application, the predicted deformation of the device satisfies the following formula:

[0027] Δx=cosJ0*Δx e +sinJ0*Δy e ;

[0028] Δy=-sinJ0*Δxe +cosJ0*Δy e ;

[0029] Δz=Δz e ;

[0030] Where J0 is the angle of the rotating joint of the rotating platform; Δx, Δy and Δz are the predicted deformations of the device; Δx e , Δy e and Δz e Predict the deformation for the end.

[0031] In the embodiment of the present application, the observation matrix satisfies the following formula:

[0032]

[0033] Among them, Δx, Δy and Δz are the predicted deformation of the equipment; x e 、y e and z e is the end position of the flexible arm in the rigid state; x0, y0 and z0 are the theoretical end positions of the flexible arm.

[0034] In the embodiment of the present application, the state transition model satisfies the following formula:

[0035]

[0036] Among them, x t-1 、y t-1 and z t-1 is the final end position of the flexible arm at time t-1; x t 、y t and z t is the predicted end position of the flexible arm at time t; v x 、v y and v z is the three-axis velocity of the end of the flexible arm at time t.

[0037] A second aspect of the present application provides a device for determining a position of an end portion of a flexible arm, comprising:

[0038] a memory configured to store instructions; and

[0039] The processor is configured to call instructions from the memory and implement the above method for determining the position of the end of the flexible arm when executing the instructions.

[0040] A third aspect of the present application provides a flexible arm mechanical device, comprising the above-mentioned device for determining the position of the end of the flexible arm.

[0041] A fourth aspect of the present application provides a machine-readable storage medium having stored thereon instructions for causing a machine to execute the above-mentioned method for determining the position of the end of a flexible arm.

[0042] Through the above technical solution, the joint angle of the target joint of the flexible arm at the current moment and the three-axis speed of the end of the flexible arm are obtained; the joint angle is input into the neural network model to obtain the predicted deformation of the end of the flexible arm; the predicted deformation of the end is coordinate-converted to obtain the predicted deformation of the device; an observation matrix is ​​established according to the predicted deformation of the device to obtain the theoretical end position; a state transition model is established according to the three-axis speed to obtain the predicted end position; Kalman filtering is performed on the theoretical end position and the predicted end position to determine the final end position of the flexible arm. The present application is based on a neural network to determine the end position of the flexible arm, obtains the three-axis speed of the end of the flexible arm through a neural network and an inertial navigation sensor, and solves the end joint position of the multi-joint flexible arm, thereby ensuring that the flexible arm mechanical equipment considers the problem of flexible arm deformation during automatic operation, improving the prediction accuracy of the end position of the flexible arm mechanical equipment, and thus improving the efficiency of automatic operation of the flexible arm mechanical equipment.

[0043] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present application but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:

[0045] Figure 1 A schematic diagram of a flow chart of a method for determining the position of an end portion of a flexible arm according to an embodiment of the present application is shown;

[0046] Figure 2 Schematically shows a structural diagram of a 7-DOF flexible arm according to an embodiment of the present application;

[0047] Figure 3 The following schematically shows a flow chart of a method for determining a final end position according to an embodiment of the present application;

[0048] Figure 4 The figure schematically shows a structural block diagram of a device for determining the position of an end of a flexible arm according to an embodiment of the present application. DETAILED DESCRIPTION

[0049] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0050] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0051] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0052] Flexible arm mechanical equipment can include construction machinery such as pump trucks and concrete placing booms. Taking a concrete placing boom as an example, in automatic concrete placing, the user only needs to set the location of the placing point, and the corresponding joint angles are obtained through an algorithm. Existing technology simplifies the arm of a flexible arm mechanical equipment into a rigid body, and the corresponding joint angles can be obtained through the rigid inverse kinematics of the arm. However, the actual arm undergoes significant elastic deformation under the influence of gravity, resulting in the end of the arm being unable to accurately reach the target placing point. Therefore, determining the position of the flexible arm end so that it can reach the desired target placing point is a very important issue in automatic concrete placing.

[0053] Figure 1 The following schematically shows a flow chart of a method for determining the position of the end of a flexible arm according to an embodiment of the present application. Figure 1 As shown, an embodiment of the present application provides a method for determining the position of the end of a flexible arm, which is applied to a flexible arm mechanical device. The method may include the following steps:

[0054] Step 101: Obtain the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment;

[0055] Step 102: Input the joint angle into the neural network model to obtain the predicted deformation of the end of the flexible arm;

[0056] Step 103: performing coordinate transformation on the predicted deformation of the terminal to obtain the predicted deformation of the device;

[0057] Step 104: Establish an observation matrix based on the predicted deformation of the equipment to obtain the theoretical end position;

[0058] Step 105: Establish a state transition model based on the three-axis speed to obtain a predicted end position;

[0059] Step 106: Perform Kalman filtering on the theoretical end position and the predicted end position to determine the final end position of the flexible arm.

[0060] Figure 2 The following schematically shows the structure of a 7-DOF flexible arm according to an embodiment of the present application. Figure 2 As shown, taking a 7-DOF flexible arm as an example, assume that A is the solved end position of the flexible arm, that is, point A shown in the figure. Due to the elastic deformation of the flexible arm, the actual end position corresponding to this set of joint angles is A', as shown by the dotted line. Therefore, how to calculate the end position of the flexible arm after deformation is an urgent problem to be solved in the embodiments of the present application.

[0061] In the embodiment of the present application, an inclination sensor or a displacement sensor can be respectively set at the target joint of the flexible arm to obtain the angle of the target joint. The processor can obtain the joint angle of the target joint through the inclination sensor or the displacement sensor set at the target joint of the flexible arm. The target joint can be part of the joint or all of the joints. An inertial navigation sensor can also be set at the end of the flexible arm. The processor can obtain the three-axis speed of the end of the flexible arm through the inertial navigation sensor set at the end of the flexible arm. Figure 2 For example, Figure 2The main influencing factor of the deformation of the flexible arm is the gravity exerted on each joint. The angle of rotation of the J0 rotation joint can be ignored, so the deformation of the flexible arm is related to J1~J6. Therefore, the target joints can be J1~J6, and the joint angles at J1~J6 are collected. At the same time, the three-axis velocity of the end of the flexible arm can be collected by the inertial navigation sensor set at the end of the flexible arm. Considering the flexible arm as rigid, the end position of the arm after the fixed rotating platform J0 can be calculated by the principle of positive kinematics, which greatly simplifies the data information that needs to be collected. Subsequently, the arm deformation information can be obtained through the total station and positive kinematics, and a database based on △x, △y and △z is established, and other motion information is collected through the inertial navigation sensor. Subsequently, 3 directions of the end of the flexible arm mechanical equipment (such as △x, △y and △z) are established according to the 3-axis deformation and 6 joint angles. Figure 2 x e 、y e and z e The neural network model of the deformation in the three directions shown.

[0062] Furthermore, the joint angles are input into the neural network model to obtain the predicted deformation of the flexible arm's end. This predicted deformation is then transformed into coordinates to obtain the predicted device deformation. An observation matrix is ​​then established based on the predicted device deformation to obtain the theoretical end position.

[0063] When large-scale mechanical equipment is in motion, the boom end experiences large vibration amplitudes and high frequencies, and the sensor acquisition frequency is low. This can easily lead to partial deviations in the collected data. Furthermore, the black-box nature of neural networks creates a certain degree of uncertainty in the resulting solution, resulting in unpredictable singularities in the output of a single network. Therefore, optimization measures are necessary to achieve more accurate and reasonable calculations of the theoretical end position. Preferably, the joint angles can be input into multiple neural network models with different structures; multiple predicted deformation components output by these multiple neural network models are obtained; the median value of these multiple predicted deformation components is selected as the reference value for a normal distribution function; the weight corresponding to each predicted deformation component is determined based on the reference value and the normal distribution function; and each predicted deformation component and its corresponding weight are weighted and squared to obtain the predicted deformation value. In this way, through the optimized combination of multiple neural networks, the information from multiple values ​​is effectively integrated, thereby filtering out singularities and jitters.

[0064] After obtaining the predicted deformation of the terminal output by the neural network, the initial position of the terminal can be obtained by combining it with the position of the rigid body. However, it is still difficult to avoid the uncertainty of the output results caused by the black box results of the neural network. Therefore, a relatively stable model is needed to adjust the entire system, and the calculation of the terminal position by inertial navigation has the problem of stable calculation in a short time but there is a cumulative error. Therefore, the embodiment of the present application proposes a method of using an inertial navigation sensor as a stable state calculation quantity and a neural network model as an observation value to correct the cumulative error. Specifically, an observation matrix is ​​established based on the predicted deformation of the device to obtain the theoretical terminal position; a state transfer model is established based on the three-axis speed to obtain the predicted terminal position. Subsequently, the extended Kalman filter algorithm is used based on the above model to assign the weights of the two model results, thereby calculating the coordinate value that integrates the inertial navigation and the neural network to determine the final end position of the flexible arm.

[0065] The embodiment of the present application proposes a method for calculating the end position of a flexible arm based on a neural network, which can solve the end joint position of a multi-joint flexible arm through a neural network Mexico and an inertial navigation sensor, ensuring that the deformation problem of the flexible arm is taken into account under automatic operation, improving the prediction accuracy of the end position of the flexible arm mechanical equipment, improving the efficiency of the automatic operation of the flexible arm mechanical equipment, and laying the foundation for the construction process of unmanned flexible arm mechanical equipment. At the same time, the embodiment of the present application can save the storage and table lookup space of the controller, occupy less resources when solving, and can improve real-time performance. In addition, the technical solution of the embodiment of the present application has a certain degree of predictability, and for the parts that are not collected, certain compensation can be performed, so that the method can continue to compensate during the operation of the equipment, improve the accuracy of the operation, and thus improve the operation accuracy of the flexible arm mechanical equipment.

[0066] In the embodiment of the present application, obtaining the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment includes:

[0067] Acquiring a joint angle of a target joint by means of an inclination sensor or a displacement sensor provided at the target joint of the flexible arm;

[0068] The three-axis velocity of the end of the flexible arm is obtained by an inertial navigation sensor arranged at the end of the flexible arm.

[0069] Specifically, an inclination sensor or a displacement sensor can be respectively set at the target joint of the flexible arm to obtain the angle of the target joint. The processor can obtain the joint angle of the target joint through the inclination sensor or the displacement sensor set at the target joint of the flexible arm. The target joint can be part of the joint or all of the joints. An inertial navigation sensor can also be set at the end of the flexible arm. The processor can obtain the three-axis speed of the end of the flexible arm through the inertial navigation sensor set at the end of the flexible arm. Figure 2For example, Figure 2 The deformation of the flexible arm in the figure is primarily influenced by the gravity acting on each joint. The rotation angle of the J0 joint is negligible, so the deformation of the flexible arm is related to joints J1 to J6. Therefore, the target joints can be J1 to J6, and the joint angles at J1 to J6 are collected. Simultaneously, the three-axis velocity of the flexible arm end can be collected using an inertial navigation sensor installed at the end of the flexible arm.

[0070] In the embodiment of the present application, inputting the joint angle into the neural network model to obtain the predicted deformation of the end of the flexible arm may include:

[0071] The joint angles are input into multiple neural network models with different structures;

[0072] Obtain multiple terminal predicted deformation components output by multiple neural network models;

[0073] The middle value of the output multiple terminal predicted deformation components is selected as the reference value of the normal distribution function;

[0074] The weight corresponding to each terminal predicted deformation component is obtained according to the reference value and the normal distribution function;

[0075] Each terminal predicted deformation component and the corresponding weight are weighted squared to obtain the terminal predicted deformation amount.

[0076] Specifically, when large-scale mechanical equipment is in motion, the vibration amplitude of the arm end is large, the vibration frequency is fast, and the sensor acquisition frequency is low, which can easily lead to partial data deviation in the collected data. In addition, due to the characteristics of the neural network black box, the solution has a certain degree of uncertainty, which causes a single network to output unpredictable singular points. Therefore, it is necessary to optimize the output results. In an embodiment of the present application, multiple neural network models with different structures can be trained, and the joint angles can be input into multiple neural network models with different structures. For example, 5 neural network models with different structures whose accuracy is within the current requirement are trained. The output results of the 5 neural network models are obtained at the same time, that is, the deformation components of multiple terminals are predicted.

[0077] The middle value of the output multiple terminal prediction deformation components is selected as the reference value of the normal distribution function. The weight corresponding to each terminal prediction deformation component is obtained according to the reference value and the normal distribution function. Then, each terminal prediction deformation component and the corresponding weight are weighted squared to obtain the terminal prediction deformation amount. For example, the middle value mid(X) of the output structure is selected as the function Parameters; then use the equation Calculate the final result of the neural network model. The principle of this equation is to use the middle value of the 5 outputs as the reference value, assign weights to each output value through the normal distribution equation, and finally solve the comprehensive value of the 5 values ​​through the above equation. After combining the normal distribution function, the equation effectively integrates the information of the 5 values. Since the normal distribution characteristics will reduce the weights of excessive values, singular points and jump points can be filtered out. In this way, through the optimization combination of multiple neural networks, the information of multiple values ​​is effectively integrated, thereby filtering out singular points and jump points. The technical solution of the embodiment of the present application can effectively reduce the generation of jump points, so that the compensation amount output by the model is more accurate and more stable.

[0078] In the embodiment of the present application, the deformation of the three axes and the multiple joint angles (such as Figure 2 J1~J6 in the figure) to establish three directions based on the end of the flexible arm mechanical device (such as Figure 2 x e 、y e and z e The neural network model of the deformation in the three directions shown in FIG. 1 can be a full neural network, and the full neural network model can satisfy the following formula:

[0079]

[0080] in, is the activation value of neuron i in layer l; is the neuron in layer l-1; For neurons The corresponding weight; is the offset.

[0081] Specifically, it can be understood as the activation value of neuron i in layer l Equal to each neuron connected to it in the previous layer Multiply by a weight Then each product is added together, plus a bias The resulting value is outputted by a function at layer l, which takes multiple joint angles as input and outputs the deformation in each direction. Since the deformation is a single value, the neural network cannot be trained using the usual right-or-wrong metric. Therefore, the difference between the output value and the training target is introduced as a model training metric.

[0082] In the embodiment of the present application, performing coordinate transformation on the terminal predicted deformation to obtain the device predicted deformation includes:

[0083] Get the angle of the rotation joint of the rotating platform at the current moment;

[0084] The predicted deformation of the terminal is converted into coordinates according to the angle to obtain the predicted deformation of the device.

[0085] Specifically, the flexible arm is considered rigid, and the end position of the arm after the fixed rotating platform J0 can be calculated by the principle of forward kinematics, which greatly simplifies the data information that needs to be collected. The predicted deformation component of the end (i.e. Figure 2 Then, based on the angle at the acquisition time J0, the predicted deformation of the equipment can be obtained by converting the model.

[0086] In the embodiment of the present application, the predicted deformation of the device satisfies the following formula:

[0087] Δx=cosJ0*Δx e +sinJ0*Δy e ;

[0088] Δy=-sinJ0*Δx e +cosJ0*Δy e ;

[0089] Δz=Δz e ;

[0090] Where J0 is the angle of the rotating joint of the rotating platform; Δx, Δy and Δz are the predicted deformations of the device; Δx e , Δy e and Δz e The deformation of the terminal is predicted. Then, the deformation of each coordinate system of the terminal can be obtained by substituting the real-time J0 angle into the above formula.

[0091] In the embodiment of the present application, the observation matrix satisfies the following formula:

[0092]

[0093] Among them, Δx, Δy and Δz are the predicted deformation of the equipment; x e 、y e and z e is the end position of the flexible arm in the rigid state; x0, y0 and z0 are the theoretical end positions of the flexible arm.

[0094] In the embodiment of the present application, the state transition model satisfies the following formula:

[0095]

[0096] Among them, x t-1 、y t-1 and z t-1 is the final end position of the flexible arm at time t-1; x t 、y t and z t is the predicted end position of the flexible arm at time t; vx 、v y and v z is the three-axis velocity of the end of the flexible arm at time t.

[0097] Specifically, after obtaining the terminal predicted deformation output by the neural network, the initial position of the terminal can be obtained in combination with the rigid body position. However, it is still difficult to avoid the uncertainty of the output results caused by the black box results of the neural network. Therefore, a relatively stable model is needed to adjust the entire system, and the terminal position calculated by inertial navigation has the problem of stable calculation in a short time but cumulative error. Therefore, the embodiment of the present application proposes a method of using an inertial navigation sensor as a stable state calculation quantity and a neural network model as an observation value to correct the cumulative error. Specifically, an observation matrix is ​​established based on the predicted deformation of the device to obtain the theoretical terminal position; a state transfer model is established based on the three-axis speed to obtain the predicted terminal position.

[0098] Then, based on the above model, the extended Kalman filter algorithm is used to assign weights to the results of the two models, thereby calculating the coordinate values ​​that integrate inertial navigation and neural network to determine the final end position of the flexible arm.

[0099] In an embodiment of the present application, Kalman filtering is performed on the theoretical end position and the predicted end position to determine the final end position of the flexible arm, including the following steps: first, the initial value of the flexible arm end position is obtained by optimizing the position compensation model of the flexible arm end, and the initial position X(0) of the system and the corresponding covariance matrix P(0) are assigned to complete the system initialization. Then, the input quantity U(k) at the current moment is updated by reading the data fed back by the inertial navigation sensor, and the current posture state is predicted once in combination with the optimal estimate of the previous moment to obtain the predicted value of the current system state. The covariance matrix of the current system is further updated by the covariance matrix of the system state at the previous moment and the predicted noise covariance Q. Then, the end position of the flexible arm obtained by the position compensation model of the optimized flexible arm end is read to obtain a new system observation posture estimation matrix, and the Kalman gain coefficient K(k) of the current system is calculated by updating the covariance matrix and observation matrix of the current system state and the noise covariance matrix R. The Kalman gain coefficient K(k) is then used to assign a proportional coefficient to the difference between the current measured value and the predicted value, further updating and optimizing the system's predicted value to obtain the system's final optimal estimate. The resulting extended Kalman gain coefficient is then used to update the covariance matrix corresponding to the current system state measurement to obtain the covariance matrix of the optimized state measurement. Finally, the optimal estimate of the system state and the corresponding covariance matrix obtained in the above steps are used as iteration parameters in the next iteration, and the second step is repeated to continuously output the system's pose estimate in real time.

[0100] Figure 3 The flowchart of the method for determining the final end position according to an embodiment of the present application is schematically shown. Figure 3 As shown, in a specific embodiment, the joint angle of the target joint of the flexible arm at the current moment and the three-axis velocity of the end of the flexible arm (i.e., inertial navigation data) are first obtained; the joint angle is input into the full neural network model to obtain the predicted deformation of the end of the flexible arm (i.e., deformation compensation value); according to the positive kinematics and the rigid end position, the predicted deformation of the end is coordinate-transformed to obtain the device predicted deformation; then an observation matrix is ​​established based on the device predicted deformation to obtain the theoretical end position (i.e., theoretical end position compensation). At the same time, a state transition model can be established based on the three-axis velocity to obtain the predicted end position; the end compensation judgment is performed based on the predicted end position and the theoretical end position, and then a Kalman filter is performed to determine the final end position of the flexible arm. This application determines the end position of the flexible arm based on a neural network, obtains the three-axis velocity of the end of the flexible arm through a neural network and an inertial navigation sensor, and solves the end joint position of the multi-joint flexible arm, ensuring that the flexible arm mechanical device considers the problem of flexible arm deformation during automatic operation, improves the prediction accuracy of the end position of the flexible arm mechanical device, and thus improves the efficiency of the automatic operation of the flexible arm mechanical device.

[0101] Figure 4 The following schematically shows a structural block diagram of a device for determining the position of the end of a flexible arm according to an embodiment of the present application. Figure 4 As shown, an embodiment of the present application provides a device for determining the position of an end of a flexible arm, which may include:

[0102] Memory 410 configured to store instructions; and

[0103] The processor 420 is configured to call instructions from the memory 410 and implement the above-mentioned method for determining the position of the end of the flexible arm when executing the instructions.

[0104] Specifically, in the embodiment of the present application, the processor 420 may be configured to:

[0105] Get the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment;

[0106] The joint angles are input into the neural network model to obtain the predicted deformation of the end of the flexible arm;

[0107] Perform coordinate transformation on the predicted deformation of the terminal to obtain the predicted deformation of the equipment;

[0108] Establish an observation matrix based on the predicted deformation of the equipment to obtain the theoretical end position;

[0109] Establish a state transfer model based on the three-axis speed to obtain the predicted end position;

[0110] The theoretical end position and the predicted end position are Kalman filtered to determine the final end position of the flexible arm.

[0111] Furthermore, the processor 420 may be further configured to:

[0112] Obtaining the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment includes:

[0113] Acquiring a joint angle of a target joint by means of an inclination sensor or a displacement sensor provided at the target joint of the flexible arm;

[0114] The three-axis velocity of the end of the flexible arm is obtained by an inertial navigation sensor arranged at the end of the flexible arm.

[0115] Furthermore, the processor 420 may be further configured to:

[0116] The joint angles are input into the neural network model to obtain the predicted deformation of the end of the flexible arm, including:

[0117] The joint angles are input into multiple neural network models with different structures;

[0118] Obtain multiple terminal predicted deformation components output by multiple neural network models;

[0119] The middle value of the output multiple terminal predicted deformation components is selected as the reference value of the normal distribution function;

[0120] The weight corresponding to each terminal predicted deformation component is obtained according to the reference value and the normal distribution function;

[0121] Each terminal predicted deformation component and the corresponding weight are weighted squared to obtain the terminal predicted deformation amount.

[0122] In the embodiment of the present application, the neural network model is a full neural network model, and the full neural network model satisfies the following formula:

[0123]

[0124] in, is the activation value of neuron i in layer l; is the neuron in layer l-1; For neurons The corresponding weight; is the offset.

[0125] Furthermore, the processor 420 may be further configured to:

[0126] The predicted deformation of the terminal is transformed into coordinates to obtain the predicted deformation of the equipment, including:

[0127] Get the angle of the rotation joint of the rotating platform at the current moment;

[0128] The predicted deformation of the terminal is converted into coordinates according to the angle to obtain the predicted deformation of the device.

[0129] In the embodiment of the present application, the predicted deformation of the device satisfies the following formula:

[0130] Δx=cosJ0*Δx e +sinJ0*Δy e ;

[0131] Δy=-sinJ0*Δx e +cosJ0*Δy e ;

[0132] Δz=Δz e ;

[0133] Where J0 is the angle of the rotating joint of the rotating platform; Δx, Δy and Δz are the predicted deformations of the device; Δx e , Δy e and Δz e Predict the deformation for the end.

[0134] In the embodiment of the present application, the observation matrix satisfies the following formula:

[0135]

[0136] Among them, Δx, Δy and Δz are the predicted deformation of the equipment; x e 、y e and z e is the end position of the flexible arm in the rigid state; x0, y0 and z0 are the theoretical end positions of the flexible arm.

[0137] In the embodiment of the present application, the state transition model satisfies the following formula:

[0138]

[0139] Among them, x t-1 、y t-1 and z t-1 is the final end position of the flexible arm at time t-1; x t 、y t and z t is the predicted end position of the flexible arm at time t; v x 、v y and v z is the three-axis velocity of the end of the flexible arm at time t.

[0140] This application determines the end position of the flexible arm based on a neural network, obtains the three-axis speed of the end of the flexible arm through the neural network and inertial navigation sensor, and solves the end joint position of the multi-joint flexible arm, ensuring that the flexible arm mechanical equipment takes into account the deformation of the flexible arm during automatic operation, and improves the prediction accuracy of the end position of the flexible arm mechanical equipment, thereby improving the efficiency of automatic operation of the flexible arm mechanical equipment.

[0141] An embodiment of the present application also provides a flexible arm mechanical device, including the above-mentioned device for determining the position of the end of the flexible arm.

[0142] An embodiment of the present application further provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the above-mentioned method for determining the position of the end of a flexible arm.

[0143] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0144] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0145] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0147] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0148] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0149] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0150] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0151] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for determining the position of an end of a flexible arm, characterized in that Applied to a flexible arm mechanical device, the method comprises: Obtaining the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment; Inputting the joint angle into a neural network model to obtain a predicted deformation of the end of the flexible arm; Performing coordinate transformation on the predicted deformation of the terminal to obtain the predicted deformation of the device; Establishing an observation matrix based on the predicted deformation of the device to obtain a theoretical end position; Establishing a state transition model based on the three-axis speed to obtain a predicted end position; Kalman filtering is performed on the theoretical end position and the predicted end position to determine the final end position of the flexible arm.

2. The method according to claim 1, characterized in that The step of obtaining the joint angle of the target joint of the flexible arm and the three-axis velocity of the end of the flexible arm at the current moment includes: Acquiring a joint angle of the target joint by means of an inclination sensor or a displacement sensor provided at the target joint of the flexible arm; The three-axis velocity of the end of the flexible arm is obtained by an inertial navigation sensor arranged at the end of the flexible arm.

3. The method according to claim 1, characterized in that Inputting the joint angle into a neural network model to obtain the predicted deformation of the end of the flexible arm includes: inputting the joint angles into a plurality of neural network models with different structures; Acquire multiple terminal predicted deformation components output by the multiple neural network models; The middle value of the output multiple terminal predicted deformation components is selected as the reference value of the normal distribution function; Obtaining a weight corresponding to each terminal predicted deformation component according to the reference value and the normal distribution function; Each terminal prediction deformation component and the corresponding weight value are weighted squared to obtain the terminal prediction deformation amount.

4. The method according to claim 1, wherein The neural network model is a full neural network model, and the full neural network model satisfies the following formula: in, is the activation value of neuron i in layer l; is the neuron in layer l-1; For neurons The corresponding weight; is the offset.

5. The method according to claim 1, wherein The step of performing coordinate transformation on the terminal predicted deformation to obtain the device predicted deformation includes: Get the angle of the rotation joint of the rotating platform at the current moment; The predicted deformation amount of the terminal is subjected to coordinate transformation according to the angle to obtain the predicted deformation amount of the device.

6. The method according to claim 5, characterized in that The predicted deformation of the device satisfies the following formula: △x=cosJ0*△x e +sinJ0*△y e ; △y=-sinJ0*△x e +cosJ0*△y e ; △z=△z e ; Wherein, J0 is the angle of the rotating joint of the rotating platform; △x, △y and △z are the predicted deformation of the device; △x e , △y e and △z e An amount of deformation is predicted for the terminal end.

7. The method according to claim 1, characterized in that The observation matrix satisfies the following formula: Among them, △x, △y and △z are the predicted deformation of the equipment; x e 、y e and z e is the end position of the flexible arm in the rigid state; x0, y0 and z0 are the theoretical end positions of the flexible arm.

8. The method according to claim 1, characterized in that The state transition model satisfies the following formula: Among them, x t-1 、y t-1 and z t-1 is the final end position of the flexible arm at time t-1; x t 、y t and z t is the predicted end position of the flexible arm at time t; v x 、v y and v z is the three-axis velocity of the end of the flexible arm at time t.

9. A device for determining the position of an end of a flexible arm, characterized in that include: a memory configured to store instructions; as well as A processor is configured to call the instructions from the memory and implement the method for determining the position of the end of a flexible arm according to any one of claims 1 to 8 when executing the instructions.

10. A flexible arm mechanical device, characterized in that: comprising a device for determining the position of the end of a flexible arm according to claim 9.

11. A machine-readable storage medium, characterized in that The machine-readable storage medium stores instructions for causing a machine to execute the method for determining the position of a distal end of a flexible arm according to any one of claims 1 to 8.

Images