WO2023000529A1

WO2023000529A1 - Robot motion analysis method and device, readable storage medium, and robot

Info

Publication number: WO2023000529A1
Application number: PCT/CN2021/126737
Authority: WO
Inventors: 白杰; 葛利刚; 陈春玉; 刘益彰; 罗秋月; 周江琛
Original assignee: 深圳市优必选科技股份有限公司
Priority date: 2021-07-20
Filing date: 2021-10-27
Publication date: 2023-01-26
Also published as: CN113618730B; CN113618730A

Abstract

A robot motion analysis method, comprising: acquiring a first joint angle of a target joint of a robot in a parallel configuration; processing the first joint angle by means of a preset forward kinematics analysis model to obtain a second joint angle of the target joint in a series configuration, the forward kinematics analysis model being a deep learning model obtained by training using a preset first training sample set, and the first training sample set being a set constructed according to an inverse kinematics analysis process. Also provided are a robot motion analysis device, a computer readable storage medium, and a robot. By means of the method, a forward kinematics analysis process is performed by means of a deep learning model; compared with existing numerical computation methods, the computational complexity is effectively reduced.

Description

A robot motion analysis method, device, readable storage medium and robot

This application claims priority to a Chinese patent application with application number 202110818134.4 filed at the China Patent Office on July 20, 2021, the entire contents of which are incorporated herein by reference.

technical field

The present application belongs to the technical field of robots, and in particular relates to a robot motion analysis method, device, computer readable storage medium and robot.

Background technique

A key point in the research of humanoid robots is the control problem of parallel configurations. For parallel configurations in robots, the analytical solution of inverse kinematics can be directly derived relatively easily through geometry or DH method, but the normal Kinematics is generally calculated by numerical method, based on the Jacobian matrix, iteratively approximated by the Newton-Raphson method, and the computational complexity is relatively high.

technical problem

In view of this, embodiments of the present application provide a robot motion analysis method, device, computer-readable storage medium, and robot to solve the problem of high computational complexity in the existing forward kinematics analysis process.

technical solution

The first aspect of the embodiment of the present application provides a robot motion analysis method, which may include:

Obtain the first joint angle of the target joint of the robot in the parallel configuration;

Use the preset positive kinematics analysis model to process the first joint angle to obtain the second joint angle of the target joint in the series configuration; the forward kinematics analysis model is the first training set by the preset The deep learning model obtained by training the sample set, and the first training sample set is a set constructed according to the inverse kinematics analysis process.

In a specific implementation of the first aspect, the robot motion analysis method may further include:

Acquiring joint motion parameters of the target joint;

Using a preset inverse dynamics analysis model to process the joint motion parameters of the target joint to obtain the driving torque of the target joint; the inverse dynamics analysis model is obtained by training a preset second training sample set A deep learning model, and the second training sample set is a set constructed according to a positive dynamics analysis process.

In a specific implementation of the first aspect, before using a preset forward kinematics analysis model to process the first joint angle, the robot motion analysis method may further include:

determining a range of motion of the target joint in a tandem configuration;

selecting a first number of serial joint angles in the range of motion;

calculating parallel joint angles corresponding to each series joint angle according to the inverse kinematics analysis process;

Constructing the first training sample set; the first training sample set includes a first number of training samples, each training sample includes a set of series joint angles and corresponding parallel joint angles;

The first training sample set is used to train the deep learning model in the initial state, and the trained deep learning model is used as the forward kinematics analysis model.

In a specific implementation of the first aspect, the deep learning model may be a generation confrontation network model including a first generator and a first discriminator;

The training of the deep learning model of the initial state using the first training sample set may include:

For each training sample in the first training sample set, use the first generator to process the parallel joint angle of the sample to obtain a first generation result;

According to the first generation result of the sample and the joint angles in series, the first discriminator is used to perform a model training process.

In a specific implementation of the first aspect, before using a preset inverse dynamics analysis model to process the joint motion parameters of the target joint, the robot motion analysis method may further include:

Acquiring the motion trajectory record of the target joint;

Selecting a second number of movement trajectory points in the movement trajectory record, wherein each movement trajectory point includes driving torque, joint velocity and joint acceleration;

Calculating joint angles corresponding to each motion track point according to the positive dynamics analysis process;

Constructing the second training sample set; the second training sample set includes a second number of training samples, each training sample includes a set of driving torque and corresponding joint motion parameters; the joint motion parameters include joint angles, joint motion parameters velocity and joint acceleration;

The deep learning model in the initial state is trained by using the second training sample set, and the trained deep learning model is used as the inverse dynamics analysis model.

In a specific implementation of the first aspect, the deep learning model is a generation confrontation network model including a second generator and a second discriminator;

The training of the deep learning model of the initial state using the second training sample set may include:

For each training sample in the second training sample set, use the second generator to process the joint motion parameters of the sample to obtain a second generation result;

According to the second generation result of the sample and the driving torque, the second discriminator is used to perform a model training process.

In a specific implementation of the first aspect, the processing of the first joint angle using a preset forward kinematics analysis model to obtain the second joint angle of the target joint in a serial configuration may include :

The first joint angle is input into the forward kinematics analysis model for processing, and the processed output of the forward kinematics analysis model is used as the second joint angle.

The second aspect of the embodiment of the present application provides a robot motion analysis device, which may include:

The first joint angle obtaining module is used to obtain the first joint angle of the target joint of the robot in the parallel configuration;

A forward kinematics analysis module, configured to use a preset forward kinematics analysis model to process the first joint angle to obtain a second joint angle of the target joint in a serial configuration; the forward kinematics analysis model It is a deep learning model trained by a preset first training sample set, and the first training sample set is a set constructed according to an inverse kinematics analysis process.

In a specific implementation of the second aspect, the robot motion analysis device may further include:

A joint motion parameter acquisition module, configured to acquire the joint motion parameters of the target joint;

The inverse dynamics analysis module is used to use a preset inverse dynamics analysis model to process the joint motion parameters of the target joint to obtain the driving torque of the target joint; the inverse dynamics analysis model is a preset The second training sample set is a deep learning model obtained by training, and the second training sample set is a set constructed according to a positive dynamics analysis process.

a range of motion determination module, configured to determine the range of motion of the target joint in a series configuration;

A serial joint angle selection module, configured to select a first number of serial joint angles in the range of motion;

an inverse kinematics analysis module, configured to calculate a parallel joint angle corresponding to each serial joint angle according to the inverse kinematics analysis process;

The first training sample set construction module is used to construct the first training sample set; the first training sample set includes a first number of training samples, and each training sample includes a set of series joint angles and corresponding parallel joint angles joint angle;

The forward kinematics analysis model training module is used to use the first training sample set to train the deep learning model in the initial state, and use the trained deep learning model as the forward kinematics analysis model.

In a specific implementation of the second aspect, the deep learning model is a generation confrontation network model including a first generator and a first discriminator;

The positive kinematics analysis model training module may include:

The first generator processing unit is configured to, for each training sample in the first training sample set, use the first generator to process the parallel joint angle of the sample to obtain a first generation result;

The first discriminator processing unit is configured to use the first discriminator to perform a model training process according to the first generation result of the sample and the series joint angle.

A movement track record acquisition module, configured to acquire the movement track record of the target joint;

A motion track point selection module, configured to select a second number of motion track points in the motion track record, wherein each motion track point includes drive torque, joint velocity and joint acceleration;

A positive dynamics analysis module, used to calculate joint angles corresponding to each motion track point according to the positive dynamics analysis process;

The second training sample set construction module is used to construct the second training sample set; the second training sample set includes a second number of training samples, and each training sample includes a set of driving torques and corresponding joint motions parameters; joint motion parameters include joint angle, joint velocity and joint acceleration;

The inverse dynamics analysis model training module is used to use the second training sample set to train the deep learning model in the initial state, and use the trained deep learning model as the inverse dynamics analysis model.

In a specific implementation of the second aspect, the deep learning model is a generation confrontation network model including a second generator and a second discriminator;

The inverse dynamics analysis model training module may include:

The second generator processing unit is configured to, for each training sample in the second training sample set, use the second generator to process the joint motion parameters of the sample to obtain a second generation result;

The second discriminator processing unit is configured to use the second discriminator to perform a model training process according to the second generation result of the sample and the driving torque.

In a specific implementation of the second aspect, the forward kinematics analysis module is specifically configured to input the first joint angle into the forward kinematics analysis model for processing, and convert the forward kinematics analysis model to The processed output is used as the second joint angle.

The third aspect of the embodiments of the present application provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any one of the above robot motion analysis methods are implemented .

The fourth aspect of the embodiments of the present application provides a robot, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, it realizes The steps of any one of the above robot motion analysis methods.

A fifth aspect of the embodiments of the present application provides a computer program product, which, when the computer program product is run on a robot, causes the robot to perform the steps of any one of the robot motion analysis methods described above.

Beneficial effect

Compared with the prior art, the embodiment of the present application has the following beneficial effects: the embodiment of the present application acquires the first joint angle of the target joint of the robot in the parallel configuration; uses the preset forward kinematics analysis model to analyze the first The joint angle is processed to obtain the second joint angle of the target joint in the series configuration; the forward kinematics analysis model is a deep learning model trained by the preset first training sample set, and the first The training sample set is a set constructed according to the inverse kinematics analysis process. Through the embodiment of the present application, the forward kinematics analysis process is performed by using the deep learning model, which effectively reduces the computational complexity compared with the existing numerical calculation method.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the accompanying drawings that need to be used in the descriptions of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings in the following description are only for the present application For some embodiments, those skilled in the art can also obtain other drawings based on these drawings without creative efforts.

Fig. 1 is the schematic diagram of the coordinate system used in the embodiment of the present application;

Fig. 2 is a corresponding relationship diagram between coordinate axes and directions of rotation;

Fig. 3 is the schematic diagram of parallel mechanism;

Fig. 4 is the schematic diagram of knee-ankle parallel mechanism;

Fig. 5 is the schematic diagram of forward kinematics and inverse kinematics analysis process;

Fig. 6 is a schematic diagram of generating an adversarial network model;

Fig. 7 is a schematic flow chart of the construction process of the forward kinematics analysis model;

Fig. 8 is a schematic flow chart of the forward kinematics analysis process;

Fig. 9 is the schematic diagram of forward kinetics and inverse kinetics analysis process;

Fig. 10 is a schematic flow chart of the construction process of the inverse dynamics analysis model;

Figure 11 is a schematic flow chart of the inverse kinetic analysis process;

FIG. 12 is a structural diagram of an embodiment of a robot motion analysis device in the embodiment of the present application;

Fig. 13 is a schematic block diagram of a robot in the embodiment of the present application.

Embodiments of the present invention

In order to make the purpose, features and advantages of the present application more obvious and understandable, 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. Obviously, the following The described embodiments are only some of the embodiments of the present application, but not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

It should be understood that when used in this specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and/or components, but does not exclude one or more other features. , whole, step, operation, element, component and/or the presence or addition of a collection thereof.

It should also be understood that the terminology used in the specification of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this specification and the appended claims, the singular forms "a", "an" and "the" are intended to include plural referents unless the context clearly dictates otherwise.

It should also be further understood that the term "and/or" used in the description of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations .

As used in this specification and the appended claims, the term "if" may be construed as "when" or "once" or "in response to determining" or "in response to detecting" depending on the context . Similarly, the phrase "if determined" or "if [the described condition or event] is detected" may be construed, depending on the context, to mean "once determined" or "in response to the determination" or "once detected [the described condition or event] ]” or “in response to detection of [described condition or event]”.

In addition, in the description of the present application, terms such as "first", "second", and "third" are only used to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

In the embodiment of the present application, a global coordinate system Σ _w as shown in FIG. 1 can be established first. In this coordinate system, the forward direction of the robot is the x-axis, the lateral direction is the y-axis, and the longitudinal direction is the z-axis.

Figure 2 shows the corresponding relationship between coordinate axes and rotation directions. As shown in the figure, the direction of rotation around the x-axis is r _x , which is recorded as the roll angle (roll angle); the direction of rotation around the y-axis is r _y , recorded as the pitch angle (pitch angle); the direction of rotation around the z axis is r _z , recorded as the yaw angle (yaw angle).

Generally, each leg of a robot may include a hip joint, a knee joint, and an ankle joint. There is a local coordinate system at each joint, and the initial state of the local coordinate system is consistent with the global coordinate system. In the series configuration, the hip joint H ₁ of the left leg and the hip joint H ₂ of the right leg have three degrees of freedom respectively, and can rotate around the x-axis, y-axis and z-axis of their local coordinate system through three rotating servos, respectively. The knee joint K ₁ of the leg and the knee joint K ₂ of the right leg have one degree of freedom respectively, which can be rotated around the y-axis of their local coordinate system through a rotating servo, and the ankle joint A ₁ of the left leg and the ankle joint A ₂ of the right leg have two degrees of freedom respectively. degrees of freedom, and can be rotated around the x-axis and y-axis of its local coordinate system through two rotating steering gears.

In the parallel configuration, a hip-knee parallel mechanism (left figure) or a knee-ankle parallel mechanism (right figure) can be set up as shown in Figure 3, where the numbers represent joint degrees of freedom, and the knee-ankle parallel mechanism is used in the following descriptions The mechanism is taken as an example, and the situation of the hip-knee parallel mechanism is similar. In the knee-ankle parallel mechanism shown in Fig. 4, the hip and knee joints are the same as the series configuration, and a local coordinate system whose initial state is consistent with the global coordinate system is established at the ankle joint (O).

Currently commonly used robot models are based on parallel configurations, but control algorithms are based on series configurations. Therefore, in the embodiment of the present application, an equivalence relationship from parallel configurations to series configurations can be established first.

Taking a single leg as an example, the joint angle in parallel configuration can be expressed as:

θ＝(θ ₁ ,θ ₂ ,θ ₃ ,θ ₄ ,θ ₅ ,θ ₆ ) ^T

Among them, θ ₁ , θ ₂ , θ ₃ are the joint angles of the hip joint in three degrees of freedom, θ ₄ is the joint angle of the knee joint in one degree of freedom, θ ₅ , θ ₆ are the joint angles of the ankle joint in two degrees of freedom joint angle on .

Correspondingly, the joint angles in the series configuration can be expressed as:

q＝(q ₁ ,q ₂ ,q ₃ ,q ₄ ,q ₅ ,q ₆ ) ^T

Among them, q ₁ , q ₂ , q ₃ are the joint angles of the hip joint in three degrees of freedom, q ₄ are the joint angles of the knee joint in one degree of freedom, q ₅ , q ₆ are the joint angles of the ankle joint in two degrees of freedom joint angle on . In the existing robot control algorithm, q can be calculated based on the waist position p _torso , R _torso and the foot position p _foot , R _foot .

In fact, for the hip joint, θ ₁ =q ₁ , θ ₂ =q ₂ , θ ₃ =q ₃ ; for the knee joint, θ ₄ =q ₄ ; and for the ankle joint, θ ₅ ,θ ₆ and q ₅ ,q ₆ is not the same.

In the embodiment of this application, as shown in Figure 5, the process of solving q ₅ and q ₆ according to θ ₅ and θ ₆ can be regarded as a forward kinematics analysis process, and the calculation of θ ₅ and θ ₆ according to q ₅ and q ₆ The process is analyzed as an inverse kinematics process.

In the embodiment of this application, the deep learning model can be used to carry out the forward kinematics analysis process. The specific type of deep learning model can be set according to the actual situation. Here, Generative Adversarial Networks (GAN) is preferably used. Model. For the GAN model, given a batch of samples, a system can be trained to generate similar samples, which can solve the problem of insufficient training data.

Figure 6 is a schematic diagram of the generation confrontational network model, which can include a generator G and a discriminator D, where the generator G is used for training and learning from a low-dimensional latent vector z(z～p _z (z) Independent and identical distribution), the mapping G(z) to the real data x; the discriminator D is used to train and learn to distinguish whether the data comes from the real data x(x~p _data (x)) or the data G(z) generated by the generator; The generation confrontation network model adjusts the generator G and the discriminator D through the optimization process, and its objective function is:

As shown in Figure 7, in a specific implementation of the embodiment of the present application, the construction process of the forward kinematics analysis model may include the following steps:

Step S701. Determine the range of motion of the target joint in the serial configuration.

Taking the target joint as the ankle joint as an example, its range of motion can be recorded as: q ₅ ∈[q _5min ,q _5max ], q ₆ ∈[q _6min ,q _6max ], where, q _5min ,q _5max ,q _6min , q _6max are preset thresholds, and their specific values can be set according to actual conditions.

Step S702, selecting a first number of serial joint angles in the range of motion.

Select a value of q ₅ and a value of q ₆ in the range of motion, and the two can form a series joint angle. The specific value of the first number can be set according to the actual situation. Generally, in order to ensure the accuracy of the model obtained through training, enough series joint angles should be collected as much as possible.

When sampling, different sampling methods can be adopted according to the actual situation, including but not limited to random sampling and uniform sampling.

Step S703 , calculating the parallel joint angle corresponding to each series joint angle according to the inverse kinematics analysis process.

For the inverse kinematics analysis process of solving the parallel joint angles θ ₅ , θ ₆ according to the series joint angles q ₅ , q ₆ , any inverse kinematics analysis method in the prior art can be selected according to the actual situation. No longer.

Step S704, constructing a first training sample set.

The first training sample set includes a first number of training samples, and each training sample includes a set of series joint angles and corresponding parallel joint angles.

Step S705, using the first training sample set to train the deep learning model in the initial state, and using the trained deep learning model as a forward kinematics analysis model.

Here, the generation confrontation network model is preferably used, and the model may include a first generator and a first discriminator. During the training process, for each training sample in the first training sample set, the first generator is used to process the parallel joint angle of the sample to obtain the first generated result, and then according to the first generated result of the sample and Connect the joint angles in series, use the first discriminator to carry out the model training process, and finally obtain the trained forward kinematics analysis model.

After training the positive kinematics analysis model, the forward kinematics analysis of the robot can be performed through the process shown in Figure 8:

Step S801, acquiring the first joint angle of the target joint of the robot in the parallel configuration.

Step S802, using the forward kinematics analysis model to process the first joint angle to obtain the second joint angle of the target joint in the serial configuration.

Specifically, the first joint angle may be input into the forward kinematics analysis model for processing, and the processed output of the forward kinematics analysis model may be used as the second joint angle.

In the embodiment of the present application, the forward kinematics analysis process is performed by using the deep learning model, which effectively reduces the computational complexity compared with the existing numerical calculation method.

As shown in Fig. 9, in a specific implementation of the embodiment of the present application, the drive torque τ of the joint, the joint speed

and joint acceleration

The process of solving the joint angle θ is regarded as a positive dynamics analysis process, which will be based on the joint angle θ, joint velocity

and joint acceleration

The process of solving the driving torque τ of the joint is regarded as the inverse dynamics analysis process.

In the embodiment of the present application, the inverse dynamics analysis process can be performed using a deep learning model. The specific type of deep learning model can be set according to the actual situation. Here, it is preferable to use a generative confrontation network model to solve the lack of training data. The problem.

As shown in Figure 10, in a specific implementation of the embodiment of the present application, the construction process of the inverse dynamics analysis model may include the following steps:

Step S1001. Obtain the motion track record of the target joint.

Step S1002, selecting a second number of motion track points in the motion track record.

Among them, each motion trajectory point includes driving torque, joint velocity and joint acceleration.

The specific value of the second number can be set according to the actual situation. Generally, in order to ensure the accuracy of the trained model, enough motion trajectory points should be collected as much as possible.

Step S1003 , calculating joint angles corresponding to each motion track point according to the forward dynamics analysis process.

For the drive torque τ, joint velocity

and joint acceleration

For the positive dynamics analysis process of solving the joint angle θ, any positive dynamics analysis method in the prior art may be selected according to the actual situation, which will not be described in this embodiment.

Step S1004, constructing a second training sample set.

The second training sample set includes a second number of training samples, and each training sample includes a set of driving torque and corresponding joint motion parameters, where the joint motion parameters include joint angle, joint velocity and joint acceleration.

Step S1005, use the second training sample set to train the deep learning model in the initial state, and use the trained deep learning model as an inverse dynamics analysis model.

Here, the generation confrontation network model is preferred, and the model may include a second generator and a second discriminator. During the training process, for each training sample in the second training sample set, first use the second generator to process the joint motion parameters of the sample to obtain the second generation result, and then according to the second generation result of the sample and Drive torque, use the second discriminator to carry out the model training process, and finally get the trained inverse dynamics analysis model.

After training the inverse dynamics analysis model, the inverse dynamics analysis of the robot can be performed through the process shown in Figure 11:

Step S1101, acquiring the joint motion parameters of the target joint.

Step S1102, using the inverse dynamics analysis model to process the joint motion parameters of the target joint to obtain the driving torque of the target joint.

Specifically, the joint motion parameters may be input into the inverse dynamics analysis model for processing, and the processed output of the inverse dynamics analysis model may be used as the driving torque.

In the embodiment of the present application, the deep learning model is used to perform the inverse dynamics analysis process, which effectively reduces the computational complexity compared with the existing numerical calculation methods.

It should be understood that the sequence numbers of the steps in the above embodiments do not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiment of the present application.

Corresponding to a robot motion analysis method described in the above embodiments, FIG. 12 shows a structural diagram of an embodiment of a robot motion analysis device provided in an embodiment of the present application.

In this embodiment, a robot motion analysis device may include:

The first joint angle obtaining module 1201 is used to obtain the first joint angle of the target joint of the robot in the parallel configuration;

The forward kinematics analysis module 1202 is configured to use a preset forward kinematics analysis model to process the first joint angle to obtain a second joint angle of the target joint in a serial configuration; the forward kinematics analysis The model is a deep learning model trained from a preset first training sample set, and the first training sample set is a set constructed according to an inverse kinematics analysis process.

In a specific implementation of the embodiment of the present application, the robot motion analysis device may further include:

The positive kinematics analysis model training module is used to use the first training sample set to train the deep learning model of the initial state, and use the trained deep learning model as the positive kinematics analysis model.

In a specific implementation of the embodiment of the present application, the deep learning model is a generative confrontation network model including a first generator and a first discriminator;

The positive kinematics analysis model training module may include:

In a specific implementation of the embodiment of the present application, the deep learning model is a generative confrontation network model including a second generator and a second discriminator;

The inverse dynamics analysis model training module may include:

In a specific implementation of the embodiment of the present application, the forward kinematics analysis module is specifically configured to input the first joint angle into the forward kinematics analysis model for processing, and convert the forward kinematics analysis The output after model processing is used as the second joint angle.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the above-described devices, modules and units can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

In the above-mentioned embodiments, the descriptions of each embodiment have their own emphases, and for parts that are not detailed or recorded in a certain embodiment, refer to the relevant descriptions of other embodiments.

FIG. 13 shows a schematic block diagram of a robot provided by the embodiment of the present application. For convenience of description, only parts related to the embodiment of the present application are shown.

As shown in FIG. 13 , the robot 13 of this embodiment includes: a processor 130 , a memory 131 , and a computer program 132 stored in the memory 131 and operable on the processor 130 . When the processor 130 executes the computer program 132, the steps in the above-mentioned embodiments of the robot motion analysis method are realized. Alternatively, when the processor 130 executes the computer program 132, the functions of the modules/units in the foregoing device embodiments are realized.

Exemplarily, the computer program 132 can be divided into one or more modules/units, and the one or more modules/units are stored in the memory 131 and executed by the processor 130 to complete this application. The one or more modules/units may be a series of computer program instruction segments capable of accomplishing specific functions, and the instruction segments are used to describe the execution process of the computer program 132 in the robot 13 .

Those skilled in the art can understand that FIG. 13 is only an example of the robot 13, and does not constitute a limitation to the robot 13. It may include more or less components than shown in the illustration, or combine certain components, or different components, such as The robot 13 may also include input and output devices, network access devices, buses, and the like.

The processor 130 can be a central processing unit (Central Processing Unit, CPU), and can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, or the like.

The storage 131 may be an internal storage unit of the robot 13 , such as a hard disk or memory of the robot 13 . The memory 131 can also be an external storage device of the robot 13, such as a plug-in hard disk equipped on the robot 13, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, Flash card (Flash Card), etc. Further, the memory 131 may also include both an internal storage unit of the robot 13 and an external storage device. The memory 131 is used to store the computer program and other programs and data required by the robot 13 . The memory 131 can also be used to temporarily store data that has been output or will be output.

Those skilled in the art can clearly understand that for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used for illustration. In practical applications, the above-mentioned functions can be assigned to different functional units, Completion of modules means that the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment may be integrated into one processing unit, or each unit may exist separately physically, or two or more units may be integrated into one unit, and the above-mentioned integrated units may adopt hardware It can also be implemented in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the present application. For the specific working process of the units and modules in the above system, reference may be made to the corresponding process in the foregoing method embodiments, and details will not be repeated here.

Those skilled in the art can appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present application.

In the embodiments provided in this application, it should be understood that the disclosed devices/robots and methods may be implemented in other ways. For example, the device/robot embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or Components may be combined or integrated into another system, or some features may be omitted, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

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

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

If the integrated module/unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments in the present application can also be completed by instructing related hardware through computer programs. The computer programs can be stored in a computer-readable storage medium, and the computer When the program is executed by the processor, the steps in the above-mentioned various method embodiments can be realized. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form. The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory) ), Random Access Memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable Storage media excludes electrical carrier signals and telecommunication signals.

The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still implement the foregoing embodiments Modifications to the technical solutions described in the examples, or equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the application, and should be included in the Within the protection scope of this application.

Claims

A method for robot motion analysis, characterized in that, comprising:

Obtain the first joint angle of the target joint of the robot in the parallel configuration;

Use the preset positive kinematics analysis model to process the first joint angle to obtain the second joint angle of the target joint in the series configuration; the forward kinematics analysis model is the first training set by the preset The deep learning model obtained by training the sample set, and the first training sample set is a set constructed according to the inverse kinematics analysis process.
The robot motion analysis method according to claim 1, further comprising:

Acquiring joint motion parameters of the target joint;

Using a preset inverse dynamics analysis model to process the joint motion parameters of the target joint to obtain the driving torque of the target joint; the inverse dynamics analysis model is obtained by training a preset second training sample set A deep learning model, and the second training sample set is a set constructed according to a positive dynamics analysis process.
The robot motion analysis method according to claim 1, further comprising: before using a preset forward kinematics analysis model to process the first joint angle:

determining a range of motion of the target joint in a tandem configuration;

selecting a first number of serial joint angles in the range of motion;

calculating parallel joint angles corresponding to each series joint angle according to the inverse kinematics analysis process;

Constructing the first training sample set; the first training sample set includes a first number of training samples, each training sample includes a set of series joint angles and corresponding parallel joint angles;

The first training sample set is used to train the deep learning model in the initial state, and the trained deep learning model is used as the forward kinematics analysis model.
The robot motion analysis method according to claim 3, wherein the deep learning model is a generation confrontation network model comprising a first generator and a first discriminator;

The training of the deep learning model of the initial state using the first training sample set includes:

For each training sample in the first training sample set, use the first generator to process the parallel joint angle of the sample to obtain a first generation result;

According to the first generation result of the sample and the joint angles in series, the first discriminator is used to perform a model training process.
The robot motion analysis method according to claim 2, wherein, before using a preset inverse dynamics analysis model to process the joint motion parameters of the target joint, further comprising:

Acquiring the motion trajectory record of the target joint;

Selecting a second number of movement trajectory points in the movement trajectory record, wherein each movement trajectory point includes driving torque, joint velocity and joint acceleration;

Calculating joint angles corresponding to each motion track point according to the positive dynamics analysis process;

Constructing the second training sample set; the second training sample set includes a second number of training samples, each training sample includes a set of driving torque and corresponding joint motion parameters; the joint motion parameters include joint angles, joint motion parameters velocity and joint acceleration;

The deep learning model in the initial state is trained by using the second training sample set, and the trained deep learning model is used as the inverse dynamics analysis model.
The robot motion analysis method according to claim 5, wherein the deep learning model is a generation confrontation network model comprising a second generator and a second discriminator;

The training of the deep learning model of the initial state using the second training sample set includes:

For each training sample in the second training sample set, use the second generator to process the joint motion parameters of the sample to obtain a second generation result;

According to the second generation result of the sample and the driving torque, the second discriminator is used to perform a model training process.
The robot motion analysis method according to any one of claims 1 to 6, characterized in that the use of the preset forward kinematics analysis model is used to process the first joint angle to obtain the target joint in series The second joint angle in configuration, including:

The first joint angle is input into the forward kinematics analysis model for processing, and the processed output of the forward kinematics analysis model is used as the second joint angle.
A robot motion analysis device is characterized in that it comprises:

The first joint angle obtaining module is used to obtain the first joint angle of the target joint of the robot in the parallel configuration;

A forward kinematics analysis module, configured to use a preset forward kinematics analysis model to process the first joint angle to obtain a second joint angle of the target joint in a serial configuration; the forward kinematics analysis model It is a deep learning model trained by a preset first training sample set, and the first training sample set is a set constructed according to an inverse kinematics analysis process.
A computer-readable storage medium, the computer-readable storage medium stores a computer program, characterized in that, when the computer program is executed by a processor, the robot motion analysis according to any one of claims 1 to 7 is realized method steps.
A robot, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, characterized in that, when the processor executes the computer program, the computer program according to claims 1 to 7 is realized. The steps of the robot motion analysis method described in any one.