A hanging rod exoskeleton upper limb motion posture prediction method and device

Abstract

The application discloses a kind of hanging rod exoskeleton upper limb movement posture prediction method and device, to solve the technical problems that exoskeleton upper limb motor response lag causes the output torque of exoskeleton shoulder, elbow joint motor and the state of person's upper limb do not match, so as to affect the compliance of operation effect and action.The present application comprises: using six-dimensional force sensor to collect the end six-dimensional force of the hanging rod exoskeleton upper limb;Using posture sensor to collect the posture information of the hanging rod exoskeleton;The end six-dimensional force and the posture information are input into the preset convolutional neural network prediction model, and the upper limb movement posture of the hanging rod exoskeleton is obtained.

Description

Technical Field

[0001] This invention relates to the field of posture prediction technology, and in particular to a method and device for predicting upper limb movement posture on a pole-mounted exoskeleton. Background Technology

[0002] When wearing an upper limb exoskeleton for pole hanging work, due to the response delay between the control signal input and the motor drive output, the upper limbs of the person wearing the exoskeleton are always in motion. When the current information is used as the control input, due to the lag in the motor response, the output torque of the exoskeleton's shoulder and elbow joint motors does not match the state of the person's upper limbs, thus affecting the work effect and the smoothness of the movement. Summary of the Invention

[0003] This invention provides a method and device for predicting the upper limb movement posture of a pole-mounted exoskeleton, which solves the technical problem that the output torque of the exoskeleton's shoulder and elbow joint motors does not match the human's upper limb state due to the lag in the response of the exoskeleton's upper limb motors, thus affecting the work performance and the smoothness of the movement.

[0004] This invention provides a method for predicting the upper limb movement posture of a pole-mounted exoskeleton, comprising:

[0005] A six-dimensional force sensor is used to collect the six-dimensional force at the distal end of the upper limb of the hanging pole exoskeleton;

[0006] An attitude sensor is used to collect the pose information of the hanging pole exoskeleton;

[0007] The six-dimensional force at the end and the pose information are input into a preset convolutional neural network prediction model to obtain the upper limb movement pose of the hanging pole exoskeleton.

[0008] Optionally, the training process of the convolutional neural network prediction model includes:

[0009] Generate an initial convolutional neural network prediction model;

[0010] The initial convolutional neural network prediction model was optimized using the Harris Eagle optimization algorithm to obtain the optimized convolutional neural network prediction model.

[0011] Acquire training sample data and the actual motion pose of each training sample data at the next moment. The training sample data includes the six-dimensional force sample of the upper limb of the hanging pole exoskeleton and the joint pose sample of the exoskeleton.

[0012] The end-effector six-dimensional force samples and the exoskeleton joint pose samples are input into the optimized convolutional neural network prediction model to generate an initial predicted pose.

[0013] By comparing the initial predicted pose with the actual motion pose, the difference data is obtained;

[0014] The optimized convolutional neural network prediction model is adjusted based on the difference data to obtain an updated convolutional neural network prediction model;

[0015] Determine if the current iteration count meets the preset iteration count;

[0016] If not, use the updated convolutional neural network prediction model as the optimized convolutional neural network prediction model, and return to the step of inputting the terminal six-dimensional force sample and the exoskeleton joint pose sample into the optimized convolutional neural network prediction model to generate the initial predicted pose;

[0017] If so, output the updated convolutional neural network prediction model as the convolutional neural network prediction model.

[0018] Optionally, the step of generating the initial convolutional neural network prediction model includes:

[0019] The number of input layer nodes is determined based on the end-effector six-dimensional force and the pose information;

[0020] The number of output layer nodes is determined based on the upper limb movement posture of the pole-mounted exoskeleton.

[0021] The number of hidden layer nodes is calculated using the number of input layer nodes and the number of output layer nodes;

[0022] The input layer is constructed using the number of nodes in the input layer, the output layer is constructed using the number of nodes in the output layer, and the hidden layer is constructed using the number of nodes in the hidden layer.

[0023] An initial convolutional neural network prediction model is constructed using the input layer, the hidden layer, and the output layer.

[0024] Optionally, the step of optimizing the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain the optimized convolutional neural network prediction model includes:

[0025] The Harris Eagle optimization algorithm is used to optimize the connection weights and thresholds of the initial convolutional neural network prediction model to obtain the optimal initial connection weights and optimal thresholds.

[0026] The initial convolutional neural network prediction model is optimized using the optimal initial connection weights and the optimal threshold to obtain the optimized convolutional neural network prediction model.

[0027] The present invention also provides a device for predicting the upper limb movement posture of a pole-mounted exoskeleton, comprising:

[0028] The end-effector six-dimensional force acquisition module is used to acquire the end-effector six-dimensional force of the upper limb of the hanging pole exoskeleton using a six-dimensional force sensor;

[0029] The pose information acquisition module is used to acquire the pose information of the hanging pole exoskeleton using a pose sensor;

[0030] The upper limb movement pose prediction module is used to input the six-dimensional force at the end and the pose information into a preset convolutional neural network prediction model to obtain the upper limb movement pose of the hanging pole exoskeleton.

[0031] Optionally, the convolutional neural network prediction model training module includes:

[0032] The initial convolutional neural network prediction model generation submodule is used to generate the initial convolutional neural network prediction model;

[0033] An optimized convolutional neural network prediction model generation submodule is used to optimize the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain an optimized convolutional neural network prediction model.

[0034] The actual motion pose acquisition submodule is used to acquire training sample data and the actual motion pose of each training sample data at the next moment. The training sample data includes the six-dimensional force sample of the upper limb of the hanging pole exoskeleton and the joint pose sample of the exoskeleton.

[0035] The initial predicted pose generation submodule is used to input the end-effector six-dimensional force sample and the exoskeleton joint pose sample into the optimized convolutional neural network prediction model to generate the initial predicted pose.

[0036] The difference data generation submodule is used to compare the initial predicted pose with the actual motion pose to obtain difference data.

[0037] An updated convolutional neural network prediction model generation submodule is used to adjust the optimized convolutional neural network prediction model based on the difference data to obtain an updated convolutional neural network prediction model;

[0038] The judgment submodule is used to determine whether the current iteration count meets the preset iteration count;

[0039] The return submodule is used to, if not, use the updated convolutional neural network prediction model as the optimized convolutional neural network prediction model, and return the step of inputting the terminal six-dimensional force sample and the exoskeleton joint pose sample into the optimized convolutional neural network prediction model to generate the initial predicted pose;

[0040] The convolutional neural network prediction model generation submodule is used to output the updated convolutional neural network prediction model as the convolutional neural network prediction model if the condition is met.

[0041] Optionally, the initial convolutional neural network prediction model generation submodule includes:

[0042] The input layer node number determination unit is used to determine the number of input layer nodes based on the end-effector six-dimensional force and the pose information.

[0043] The output layer node number determination unit is used to determine the number of output layer nodes based on the upper limb movement posture of the hanging pole exoskeleton.

[0044] A hidden layer node count calculation unit is used to calculate the number of hidden layer nodes using the number of input layer nodes and the number of output layer nodes;

[0045] A hierarchical building unit is used to build an input layer using the number of input layer nodes, an output layer using the number of output layer nodes, and a hidden layer using the number of hidden layer nodes.

[0046] An initial convolutional neural network prediction model building unit is used to construct an initial convolutional neural network prediction model using the input layer, the hidden layer, and the output layer.

[0047] Optionally, the optimized convolutional neural network prediction model generation submodule includes:

[0048] The optimal initial connection weights and optimal thresholds acquisition unit is used to optimize the connection weights and thresholds of the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain the optimal initial connection weights and optimal thresholds.

[0049] An optimized convolutional neural network prediction model generation unit is used to optimize the initial convolutional neural network prediction model using the optimal initial connection weights and the optimal threshold to obtain an optimized convolutional neural network prediction model.

[0050] The present invention also provides an electronic device, the device comprising a processor and a memory:

[0051] The memory is used to store program code and transmit the program code to the processor;

[0052] The processor is used to execute the upper limb motion pose prediction method of the pole-mounted exoskeleton as described above, according to the instructions in the program code.

[0053] The present invention also provides a computer-readable storage medium for storing program code for executing the upper limb motion pose prediction method of the pole-mounted exoskeleton as described in any of the preceding claims.

[0054] As can be seen from the above technical solution, the present invention has the following advantages: The present invention uses a six-dimensional force sensor to collect the six-dimensional force at the end of the upper limb of the pole-mounted exoskeleton; it uses a posture sensor to collect the pose information of the pole-mounted exoskeleton; and it inputs the six-dimensional force and pose information into a preset convolutional neural network prediction model to obtain the motion pose of the upper limb of the pole-mounted exoskeleton. Thus, by responding in advance with the predicted pose information for the next moment, the influence of the mismatch between the output torque of the exoskeleton's shoulder and elbow joint motors and the human upper limb state is eliminated, improving work efficiency and the smoothness of movement. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 A flowchart illustrating the steps of a method for predicting upper limb movement posture using a pole-mounted exoskeleton, as provided in an embodiment of the present invention.

[0057] Figure 2 This is a schematic diagram illustrating the basic working principle of a convolutional neural network.

[0058] Figure 3 A flowchart illustrating the steps of a convolutional neural network prediction model training method provided in an embodiment of the present invention;

[0059] Figure 4 This is a schematic diagram of the HHO algorithm optimization process;

[0060] Figure 5 This is a structural block diagram of an upper limb movement posture prediction device for a pole-mounted exoskeleton provided in an embodiment of the present invention. Detailed Implementation

[0061] This invention provides a method and device for predicting the upper limb movement posture of a pole-mounted exoskeleton, which solves the technical problem that the output torque of the exoskeleton's shoulder and elbow joint motors does not match the human's upper limb state due to the lag in the response of the exoskeleton's upper limb motors, thus affecting the work performance and the smoothness of the movement.

[0062] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0063] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a method for predicting upper limb movement posture using a pole-mounted exoskeleton, as provided in an embodiment of the present invention.

[0064] The present invention provides a method for predicting the upper limb movement posture of a pole-mounted exoskeleton, which may specifically include the following steps:

[0065] Step 101: Use a six-dimensional force sensor to collect the six-dimensional force at the end of the upper limb of the hanging pole exoskeleton;

[0066] An exoskeleton is a mechanical device made of a steel frame that a person can wear. This equipment can provide extra energy for the movement of the limbs.

[0067] A six-dimensional force sensor is a sensor used to measure the forces and torques acting on an object. It can measure the forces acting on an object in six degrees of freedom, including forces in three directions and torques in three directions.

[0068] In this embodiment of the invention, a six-dimensional force sensor can be installed at the end of each joint of the upper limb of the pole exoskeleton to collect the six-dimensional force of each joint of the upper limb of the pole exoskeleton.

[0069] Step 102: Use an attitude sensor to collect the pose information of the pole exoskeleton;

[0070] The attitude sensor (ET-ahrs) is a high-performance three-dimensional motion attitude measurement system based on MEMS technology. It includes auxiliary motion sensors such as a three-axis gyroscope, a three-axis accelerometer (i.e., IMU), and a three-axis electronic compass. It outputs calibrated angular velocity, acceleration, magnetic data, etc. through an embedded low-power ARM processor, and performs motion attitude measurement through a quaternion-based sensor data algorithm, outputting zero-drift three-dimensional attitude data in real time expressed in quaternions, Euler angles, etc.

[0071] In this embodiment of the invention, posture sensors can be installed at each joint of the upper limb of the pole-mounted exoskeleton to collect the position and posture information of each joint of the upper limb of the pole-mounted exoskeleton.

[0072] In one example, pose information may include the shoulder and elbow poses of the pole-mounted exoskeleton.

[0073] Step 103: Input the six-dimensional force and pose information of the end effector into the preset convolutional neural network prediction model to obtain the upper limb movement pose of the pole exoskeleton.

[0074] Convolutional Neural Networks (CNNs) are a type of feedforward neural network that incorporates convolutional computations and has a deep structure. Their basic working principle is as follows: Figure 2 As shown, the network consists of several parts, including an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. Its training and learning process mainly consists of two processes: forward propagation and backward propagation. During forward propagation, the features of the sample data are input into the input layer, processed by each hidden layer, and finally output from the output layer. Backpropagation, on the other hand, propagates the error signal between the actual output and the expected output of the network from the output layer back to the input layer, thereby obtaining the error learning signal of each layer. Based on the error learning signal, the weights and thresholds of the neurons in each layer are adjusted. Through repeated iterative learning, the actual output gradually approaches the expected output, minimizing the error and achieving the accuracy of the learning target.

[0075] In this embodiment of the invention, the six-dimensional force and pose information at the end of the hanging pole exoskeleton are input into the convolutional neural network prediction model, and the vector representation is as follows:

[0076]

[0077] The output is the upper limb motion pose of the pole-mounted exoskeleton, including the predicted poses of the shoulder and elbow joints:

[0078]

[0079] During forward propagation computation in the network, nodes at different layers are connected and passed between each other via weights. Finally, the output value is obtained through an activation function, and the calculation formula for the output can be expressed as follows:

[0080]

[0081] Where n represents the number of upper-level nodes, ω i Indicates the connection weight, x i Let f(x) represent the output of the upper-level node, θ represent the threshold, and f(x) be the activation function, using the sigmoid function, expressed in the following formula:

[0082]

[0083] During backpropagation, the mean square error of the network output is calculated using the following formula:

[0084]

[0085] Where l is the number of training samples. y is the predicted value output by the network. i Let the desired output value be denoted as . Then, the error signal of each layer is calculated by backpropagation. Based on the gradient descent strategy, and with a suitable learning rate set, the network parameters are adjusted, and the weight update formula is as follows:

[0086] Δω j =ηδ j y i

[0087] Where, Δω ij δ represents the weight update amount between the i-th node and the j-th node; j y represents the error signal of the j-th node; i This represents the output of the i-th node; η is the learning rate, which determines the magnitude of weight updates. If η is too large, the network will fail to converge; if η is too small, the convergence speed will be slow. The learning rate is set between 0.01 and 0.8.

[0088] This invention employs a six-dimensional force sensor to collect the six-dimensional force at the end of the upper limb of a pole-mounted exoskeleton; and a posture sensor to collect the posture information of the exoskeleton. The six-dimensional force and posture information are then input into a pre-defined convolutional neural network prediction model to obtain the movement posture of the upper limb of the pole-mounted exoskeleton. By responding in advance based on the predicted posture information for the next moment, the invention eliminates the influence of the mismatch between the output torque of the exoskeleton's shoulder and elbow joint motors and the user's upper limb state, thereby improving work efficiency and the smoothness of movement.

[0089] Please see Figure 3 , Figure 3 A flowchart illustrating the steps of a convolutional neural network prediction model training method provided in an embodiment of the present invention.

[0090] Step 201: Generate the initial convolutional neural network prediction model;

[0091] In one example, the steps to generate an initial convolutional neural network prediction model may specifically include the following sub-steps:

[0092] S11, determine the number of input layer nodes based on the end-effector six-dimensional force and pose information;

[0093] S12, determine the number of output layer nodes based on the upper limb movement posture of the pole-mounted exoskeleton;

[0094] S13, calculate the number of hidden layer nodes using the number of input layer nodes and the number of output layer nodes;

[0095] S14, construct the input layer using the number of nodes in the input layer, construct the output layer using the number of nodes in the output layer, and construct the hidden layer using the number of nodes in the hidden layer;

[0096] S15 uses an input layer, a hidden layer, and an output layer to construct an initial convolutional neural network prediction model.

[0097] In the specific implementation, an initial convolutional neural network prediction model can be built, consisting of an input layer, hidden layers, and an output layer. The input layer has 14 nodes, corresponding to the feature vectors of each component of the six-dimensional force sensor and each component of the pose sensor; the output layer has 8 nodes, corresponding to the shoulder and elbow joint pose outputs of the exoskeleton; the number of nodes in the hidden layer can be initially obtained based on an empirical formula, as shown below:

[0098]

[0099] Where m is the number of input layer nodes, n is the number of output layer nodes, and a is a constant, usually taken as a∈(1, 12). According to preliminary calculations, the number of hidden layer nodes is approximately in the range of h∈[5, 16].

[0100] Step 202: The Harris Eagle optimization algorithm is used to optimize the initial convolutional neural network prediction model to obtain the optimized convolutional neural network prediction model.

[0101] The Harris Eagle Optimization Algorithm (HHO) is a metaheuristic algorithm that simulates the cooperative foraging behavior of Harris Eagles to find the optimal solution. Convolutional neural networks (CNNs) can be trained and fitted using historical data, exhibiting strong nonlinear fitting capabilities. Combining the HHO optimization algorithm and CNNs effectively improves the model's performance and stability, yielding optimal prediction results.

[0102] like Figure 4 As shown, the HHO algorithm optimization process includes the following steps:

[0103] 1. Set algorithm parameters and initialize the population;

[0104] 2. Calculate the fitness value and determine the optimal individual;

[0105] 3. Update the prey escape energy factor E and distance J;

[0106] 4. Determine if |E| > 1. If yes, update the individual position and proceed to step 8; otherwise, proceed to step 5.

[0107] 5. Determine if |E| < 0.5. If yes, proceed to step 6; otherwise, proceed to step 7.

[0108] 6. Determine if R ≥ 0.5. If yes, adopt a strong encirclement strategy and jump to step 8; otherwise, adopt a gradual dive strong encirclement strategy and jump to step 8.

[0109] 7. Determine if R < 0.5. If yes, adopt a gradual dive soft encirclement strategy and jump to step 8; otherwise, adopt a soft encirclement strategy and jump to step 8.

[0110] 8. Determine if T > tmax; if not, go to step 3; if yes, end the optimization process.

[0111] In one example, the Harris Eagle optimization algorithm is used to optimize the initial convolutional neural network prediction model, resulting in the steps for optimizing the convolutional neural network prediction model. These steps may include the following sub-steps:

[0112] S21, the Harris Eagle optimization algorithm is used to optimize the connection weights and thresholds of the initial convolutional neural network prediction model to obtain the optimal initial connection weights and optimal thresholds;

[0113] S22, the initial convolutional neural network prediction model is optimized using the optimal initial connection weights and the optimal threshold to obtain the optimized convolutional neural network prediction model.

[0114] In practical implementation, the HHO algorithm can be used to find the optimal initial connection weights and optimal threshold of the initial convolutional neural network. First, in the exploration phase, the next time step position is generated based on the random position and the positions of other hawks. Then, the process transitions to the development phase, where a soft or hard siege strategy is selected based on the escape probability. This process is iterated repeatedly until the maximum number of iterations is reached or the error requirement is met, outputting the optimal initial connection weights and optimal threshold. Finally, the optimal initial connection weights and optimal threshold are used to optimize the initial convolutional neural network prediction model, resulting in the optimized convolutional neural network prediction model.

[0115] Step 203: Obtain training sample data and the actual motion pose of each training sample data at the next moment. The training sample data includes the six-dimensional force samples of the upper limb of the hanging pole exoskeleton and the joint pose samples of the exoskeleton.

[0116] Step 204: Input the end-effector six-dimensional force samples and exoskeleton joint pose samples into the optimized convolutional neural network prediction model to generate the initial predicted pose;

[0117] Step 205: Compare the initial predicted pose with the actual motion pose to obtain the difference data;

[0118] Step 206: Adjust and optimize the convolutional neural network prediction model based on the difference data to obtain the updated convolutional neural network prediction model;

[0119] Step 207: Determine whether the current iteration count meets the preset iteration count;

[0120] Step 208: If not, update the convolutional neural network prediction model as the optimized convolutional neural network prediction model, and return to the step of inputting the end six-dimensional force samples and exoskeleton joint pose samples into the optimized convolutional neural network prediction model to generate the initial predicted pose.

[0121] Step 209: If yes, output the updated convolutional neural network prediction model as the convolutional neural network prediction model.

[0122] After generating the optimized convolutional neural network prediction model, training sample data and the actual motion pose of each training sample data at the next moment can be obtained. The training sample data includes the six-dimensional force samples of the upper limb of the pole-mounted exoskeleton and the exoskeleton joint pose samples. Then, the six-dimensional force samples of the upper limb and the exoskeleton joint pose samples are input into the optimized convolutional neural network prediction model to generate the initial predicted pose. Next, the initial predicted pose is compared with the actual motion pose to obtain the difference data. Finally, the optimized convolutional neural network prediction model is adjusted according to the difference data to obtain the updated convolutional neural network prediction model. It is then determined whether the current iteration number meets the preset iteration number. If not, the updated convolutional neural network prediction model is used as the optimized convolutional neural network prediction model, and the step of inputting the six-dimensional force samples of the upper limb and the exoskeleton joint pose samples into the optimized convolutional neural network prediction model to generate the initial predicted pose is returned. If yes, the updated convolutional neural network prediction model is output as the convolutional neural network prediction model.

[0123] Optimizing convolutional neural network (CNN) prediction models using the HHO algorithm can eliminate problems such as susceptibility to local minima, sensitivity to initial parameters, significant influence from network structure parameters, and slow training speed. This results in improved prediction performance for CNN models.

[0124] Please see Figure 5 , Figure 5 This is a structural block diagram of an upper limb movement posture prediction device for a pole-mounted exoskeleton provided in an embodiment of the present invention.

[0125] This invention provides a device for predicting the upper limb movement posture of a pole-mounted exoskeleton, comprising:

[0126] The six-dimensional force acquisition module 501 is used to acquire the six-dimensional force at the end of the upper limb of the pole exoskeleton using a six-dimensional force sensor;

[0127] The pose information acquisition module 502 is used to acquire pose information of the hanging pole exoskeleton using a pose sensor;

[0128] The upper limb movement pose prediction module 503 is used to input the six-dimensional force and pose information of the end effector into a preset convolutional neural network prediction model to obtain the upper limb movement pose of the pole exoskeleton.

[0129] In this embodiment of the invention, the convolutional neural network prediction model training module includes:

[0130] The initial convolutional neural network prediction model generation submodule is used to generate the initial convolutional neural network prediction model;

[0131] The optimized convolutional neural network prediction model generation submodule is used to optimize the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain the optimized convolutional neural network prediction model.

[0132] The actual motion pose acquisition submodule is used to acquire training sample data and the actual motion pose of each training sample data at the next moment. The training sample data includes the six-dimensional force samples of the upper limb of the pole-mounted exoskeleton and the joint pose samples of the exoskeleton.

[0133] The initial predicted pose generation submodule is used to input the end-effector six-dimensional force samples and exoskeleton joint pose samples into the optimized convolutional neural network prediction model to generate the initial predicted pose;

[0134] The difference data generation submodule is used to compare the initial predicted pose with the actual motion pose to obtain difference data.

[0135] The updated convolutional neural network prediction model generation submodule is used to adjust and optimize the convolutional neural network prediction model based on the difference data, and obtain the updated convolutional neural network prediction model;

[0136] The judgment submodule is used to determine whether the current iteration count meets the preset iteration count;

[0137] Return submodule, which, if not, updates the convolutional neural network prediction model as the optimized convolutional neural network prediction model, and returns the steps of inputting the end six-dimensional force samples and exoskeleton joint pose samples into the optimized convolutional neural network prediction model to generate the initial predicted pose;

[0138] The convolutional neural network prediction model generation submodule is used to update the convolutional neural network prediction model if the condition is met.

[0139] In this embodiment of the invention, the initial convolutional neural network prediction model generation submodule includes:

[0140] The input layer node number determination unit is used to determine the number of input layer nodes based on the end-effector six-dimensional force and pose information.

[0141] The output layer node number determination unit is used to determine the number of output layer nodes based on the upper limb movement posture of the pole-mounted exoskeleton.

[0142] The hidden layer node count calculation unit is used to calculate the number of hidden layer nodes using the number of input layer nodes and the number of output layer nodes.

[0143] Hierarchical building units are used to construct input layers based on the number of input layer nodes, output layers based on the number of output layer nodes, and hidden layers based on the number of hidden layer nodes.

[0144] The initial convolutional neural network prediction model building unit is used to construct an initial convolutional neural network prediction model using an input layer, a hidden layer, and an output layer.

[0145] In this embodiment of the invention, the optimized convolutional neural network prediction model generation submodule includes:

[0146] The optimal initial connection weights and optimal thresholds acquisition unit is used to optimize the connection weights and thresholds of the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain the optimal initial connection weights and optimal thresholds.

[0147] An optimized convolutional neural network prediction model generation unit is used to optimize the initial convolutional neural network prediction model using the optimal initial connection weights and the optimal threshold, thereby obtaining an optimized convolutional neural network prediction model.

[0148] This invention also provides an electronic device, which includes a processor and a memory:

[0149] The memory is used to store program code and transfer the program code to the processor;

[0150] The processor is used to execute the upper limb motion pose prediction method of the pole-mounted exoskeleton according to the instructions in the program code.

[0151] This invention also provides a computer-readable storage medium for storing program code for executing the upper limb motion pose prediction method of the pole-mounted exoskeleton according to this invention.

[0152] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0153] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0154] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0155] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0156] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0157] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0158] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0159] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0160] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the upper limb movement posture of a pole-mounted exoskeleton, characterized in that, include: A six-dimensional force sensor is used to collect the six-dimensional force at the distal end of the upper limb of the hanging pole exoskeleton; An attitude sensor is used to collect the pose information of the hanging pole exoskeleton; The six-dimensional force at the end and the pose information are input into a preset convolutional neural network prediction model to obtain the upper limb movement pose of the hanging pole exoskeleton; The training process of the convolutional neural network prediction model includes: Generate an initial convolutional neural network prediction model; The initial convolutional neural network prediction model was optimized using the Harris Eagle optimization algorithm to obtain the optimized convolutional neural network prediction model. Acquire training sample data and the actual motion pose of each training sample data at the next moment. The training sample data includes the six-dimensional force sample of the upper limb of the hanging pole exoskeleton and the joint pose sample of the exoskeleton. The end-effector six-dimensional force samples and the exoskeleton joint pose samples are input into the optimized convolutional neural network prediction model to generate an initial predicted pose. By comparing the initial predicted pose with the actual motion pose, the difference data is obtained; The optimized convolutional neural network prediction model is adjusted based on the difference data to obtain an updated convolutional neural network prediction model; Determine if the current iteration count meets the preset iteration count; If not, use the updated convolutional neural network prediction model as the optimized convolutional neural network prediction model, and return to the step of inputting the terminal six-dimensional force sample and the exoskeleton joint pose sample into the optimized convolutional neural network prediction model to generate the initial predicted pose; If so, output the updated convolutional neural network prediction model as the convolutional neural network prediction model.

2. The method according to claim 1, characterized in that, The step of generating the initial convolutional neural network prediction model includes: The number of input layer nodes is determined based on the end-effector six-dimensional force and the pose information; The number of output layer nodes is determined based on the upper limb movement posture of the pole-mounted exoskeleton. The number of hidden layer nodes is calculated using the number of input layer nodes and the number of output layer nodes; The input layer is constructed using the number of nodes in the input layer, the output layer is constructed using the number of nodes in the output layer, and the hidden layer is constructed using the number of nodes in the hidden layer. An initial convolutional neural network prediction model is constructed using the input layer, the hidden layer, and the output layer.

3. The method according to claim 1, characterized in that, The step of optimizing the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain the optimized convolutional neural network prediction model includes: The Harris Eagle optimization algorithm is used to optimize the connection weights and thresholds of the initial convolutional neural network prediction model to obtain the optimal initial connection weights and optimal thresholds. The initial convolutional neural network prediction model is optimized using the optimal initial connection weights and the optimal threshold to obtain the optimized convolutional neural network prediction model.

4. A device for predicting the upper limb movement posture of a pole-mounted exoskeleton, characterized in that, include: The end-effector six-dimensional force acquisition module is used to acquire the end-effector six-dimensional force of the upper limb of the hanging pole exoskeleton using a six-dimensional force sensor; The pose information acquisition module is used to acquire the pose information of the hanging pole exoskeleton using a pose sensor; The upper limb movement pose prediction module is used to input the six-dimensional force at the end and the pose information into a preset convolutional neural network prediction model to obtain the upper limb movement pose of the hanging pole exoskeleton. The convolutional neural network prediction model training module includes: The initial convolutional neural network prediction model generation submodule is used to generate the initial convolutional neural network prediction model; An optimized convolutional neural network prediction model generation submodule is used to optimize the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain an optimized convolutional neural network prediction model. The actual motion pose acquisition submodule is used to acquire training sample data and the actual motion pose of each training sample data at the next moment. The training sample data includes the six-dimensional force sample of the upper limb of the hanging pole exoskeleton and the joint pose sample of the exoskeleton. The initial predicted pose generation submodule is used to input the end-effector six-dimensional force sample and the exoskeleton joint pose sample into the optimized convolutional neural network prediction model to generate the initial predicted pose. The difference data generation submodule is used to compare the initial predicted pose with the actual motion pose to obtain difference data. An updated convolutional neural network prediction model generation submodule is used to adjust the optimized convolutional neural network prediction model based on the difference data to obtain an updated convolutional neural network prediction model; The judgment submodule is used to determine whether the current iteration count meets the preset iteration count; The return submodule is used to, if not, use the updated convolutional neural network prediction model as the optimized convolutional neural network prediction model, and return the step of inputting the terminal six-dimensional force sample and the exoskeleton joint pose sample into the optimized convolutional neural network prediction model to generate the initial predicted pose; The convolutional neural network prediction model generation submodule is used to output the updated convolutional neural network prediction model as the convolutional neural network prediction model if the condition is met.

5. The apparatus according to claim 4, characterized in that, The initial convolutional neural network prediction model generation submodule includes: The input layer node number determination unit is used to determine the number of input layer nodes based on the end-effector six-dimensional force and the pose information. The output layer node number determination unit is used to determine the number of output layer nodes based on the upper limb movement posture of the hanging pole exoskeleton. A hidden layer node count calculation unit is used to calculate the number of hidden layer nodes using the number of input layer nodes and the number of output layer nodes; A hierarchical building unit is used to build an input layer using the number of input layer nodes, an output layer using the number of output layer nodes, and a hidden layer using the number of hidden layer nodes. An initial convolutional neural network prediction model building unit is used to construct an initial convolutional neural network prediction model using the input layer, the hidden layer, and the output layer.

6. The apparatus according to claim 4, characterized in that, The optimized convolutional neural network prediction model generation submodule includes: The optimal initial connection weights and optimal thresholds acquisition unit is used to optimize the connection weights and thresholds of the initial convolutional neural network prediction model using the Harris Eagle optimization algorithm to obtain the optimal initial connection weights and optimal thresholds. An optimized convolutional neural network prediction model generation unit is used to optimize the initial convolutional neural network prediction model using the optimal initial connection weights and the optimal threshold to obtain an optimized convolutional neural network prediction model.

7. An electronic device, characterized in that, The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the upper limb motion posture prediction method of the pole-mounted exoskeleton according to any one of the instructions in the program code.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program code, which is used to execute the upper limb motion posture prediction method of the pole-mounted exoskeleton according to any one of claims 1-3.

Images