Robotic arm learning method and apparatus for anthropomorphizing arm stiffness and motor skills

Abstract

The present application relates to the technical field of robot control and machine learning, and specifically provides a humanoid arm stiffness and motion skill mechanical arm learning method, computer equipment and medium, including teaching demonstration and learning two stages, the teaching demonstration stage collects task trajectory information and human arm stiffness information, the learning stage completes the extraction of the trajectory information and the arm stiffness information and establishes the corresponding motion trajectory model and the adaptive stiffness model, realizes the control to the mechanical arm, completes the motion planning to the task trajectory while learning the humanoid stiffness skill, lets the mechanical arm have the motion skill while having the variable stiffness skill, can let the mechanical arm be applied to more working occasions such as man-machine cooperation, auxiliary medical treatment and the like.

Description

Technical Field

[0001] This invention relates to the fields of robot control and machine learning technology, and in particular to a method, device and medium for learning the stiffness and motion skills of a human-like arm. Background Technology

[0002] With the widespread application of robots, people have increasingly higher performance requirements for them, hoping they can possess more human-like attributes. For example, most robots currently only possess human-like movement skills, capable only of point-to-point movement or movement along fixed trajectories. The variable stiffness characteristic of the human arm can help humans complete various tasks. However, research on robot stiffness skill learning algorithms is currently scarce; most studies pre-set robot stiffness. Unlike a human arm, a robot cannot flexibly change its stiffness according to actual working conditions to complete tasks. In many fields such as multi-robot collaboration, human-robot collaboration, exoskeleton assistance, and medical rehabilitation, robots need to possess stiffness skills similar to those of a human arm. Therefore, it is essential for robots to learn and master human-like stiffness skills.

[0003] Demonstration learning has proven to be an effective skill learning method for robots. By learning relevant skills from human demonstrations, robots fully utilize human operational experience and then generalize the learned skills to apply them to new tasks, enabling robots to achieve human-like operational capabilities. Compared to traditional programming-controlled robots, demonstration learning is more efficient and flexible.

[0004] In conclusion, how to enable robots to learn human-like stiffness skills is a problem that needs to be solved. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method, device, and medium for learning the stiffness and motor skills of a robotic arm, similar to those of a human-shaped arm.

[0006] In a first aspect, embodiments of the present invention provide a method for learning the stiffness and motor skills of a robotic arm, comprising:

[0007] An electromyography (EMG) sensor is worn on the arm of the instructor, and the EMG sensor is used to test the stiffness information of the instructor's arm.

[0008] A force sensor and a workload are installed at the end of a robotic arm. The force sensor is used to measure the weight of the workload to obtain workload weight information.

[0009] The demonstrator drags the mechanical arm from a demonstration starting position to a demonstration target position, and obtains trajectory information and arm rigidity information of the mechanical arm during the dragging process, wherein the trajectory information comprises time series and position information of an end in three directions of XYZ;

[0010] The trajectory information is preprocessed, and velocity and acceleration information of each direction at any time is determined according to the position information and the time series;

[0011] The motion trajectory of the mechanical arm is learned by using a DMP algorithm combined with the velocity and acceleration information, to complete learning of a motion trajectory model in the demonstration stage;

[0012] The time series and the arm rigidity information are fitted by using a Gaussian kernel function, and a regular system is used for driving, and a local weighted regression algorithm is used for learning a weight parameter, to complete learning of a rigidity fitting term in the demonstration process;

[0013] An adaptive coefficient is combined with the rigidity fitting term to obtain an adaptive rigidity model in the demonstration stage;

[0014] Task information under a new task is obtained, the task information comprising a starting position, a target position and a motion time, the motion trajectory model outputting a predicted motion trajectory based on the task information, and the adaptive rigidity model outputting predicted rigidity control information based on the task information;

[0015] The mechanical arm is controlled to work based on the predicted motion trajectory and the predicted rigidity control information.

[0016] As an optional solution, the learning of the motion trajectory of the mechanical arm by using the DMP algorithm combined with the velocity and acceleration information, to complete learning of the motion trajectory model in the demonstration stage, comprises:

[0017] The mathematical model of the DMP algorithm is as follows:

[0018] In the formula, represents a constant coefficient; represents a constant coefficient; represents a target position; represents a position at any time; represents a velocity at any time; represents an acceleration at any time; represents a system phase; represents a time proportion coefficient;

[0019] The part before the plus sign in the formula is a second-order stable convergence system, which ensures that the mechanical arm can reach the target point, and in order to make the second-order system respond faster and reach the target point, the following conditions need to be met Let the system be in a critical damping state, wherein The running time of the system after learning can be adjusted, when The running time of the system is shorter than the time of the teaching process, when The running time of the system is longer than the time of the teaching process, when

[0020] The part after the plus sign is a nonlinear forcing term, which is used to ensure that the mechanical arm can move in the shape of the trajectory in the teaching process, and to retain the characteristic information of the movement trajectory in the teaching process, and the nonlinear forcing term is as follows:

[0021]

[0022] In the formula, Indicates a Gaussian kernel function; Indicates the number of Gaussian kernel functions; Indicates the weight of each Gaussian kernel function; Indicates the starting position; Indicates the center of the Gaussian kernel function; Indicates the variance of the Gaussian kernel function; By determining the number and weight parameters of the Gaussian kernel function, the characteristics of the movement trajectory in the teaching process are learned;

[0023] Where the system phase of the regular system Satisfies , Is a constant and satisfies , wherein Is a function of time t;

[0024] The position information, velocity and acceleration information of the mechanical arm in XYZ three directions at any time are obtained, the constant in the mathematical model of the DMP is set, and the value of each time is determined according to the mathematical model of the DMP , which is ;

[0025] Based on the obtained series of values, the weight of each Gaussian kernel function is learned, so that The value is close to the series of values obtained by the movement trajectory in the teaching process, wherein the determination of the number of Gaussian kernel functions, the more the number is, the higher the fitting accuracy is, and the error value is minimized

[0026] The local weighted regression algorithm is adopted to learn the weight parameter, and the weight of the i-th Gaussian kernel function is obtained , wherein:

[0027]

[0028] The determination of the weight parameters in the model is completed by the above formula, the learning of the motion trajectory model in the teaching stage is completed, the motion trajectory model is used to generate a motion trajectory under the driving of the system phase according to the start position, the target position and the running time set according to the task information of the mechanical arm.

[0029] As an optional solution, the time series and the arm stiffness information are fitted by using the Gaussian kernel function, and the weight parameters are learned by using a local weighted regression algorithm, the learning of the stiffness fitting term in the teaching process is completed, including:

[0030] N Gaussian kernel functions are used to fit the stiffness curve in the teaching process, and a regular system is used for driving, which is represented as: According to the arm stiffness information collected at each moment in the teaching process, the sequence is obtained, and the error is minimized, and the weight parameters are learned by using a local weighted regression algorithm, and the solution is: In the formula:

[0031]

[0032] The determination of the weight parameters is completed by the above formula, and the learning of the stiffness fitting term in the teaching process is completed.

[0033] As an optional solution, the adaptive coefficient is combined with the stiffness fitting term to obtain a teaching stage adaptive stiffness model, including:

[0034] The adaptive coefficient is combined with the stiffness fitting term to obtain a teaching stage adaptive stiffness model, and the expression of the adaptive coefficient is: In the formula: represents actual working load weight information; represents teaching working load weight information; and a formula of the adaptive stiffness model is as follows: .

[0035] In a second aspect, an embodiment of the present application provides a computer device, including:

[0036] at least one processor; and

[0037] a memory in communication connection with the at least one processor; wherein

[0038] ​The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above-mentioned human-like arm stiffness and motor skill learning method of the robot arm.

[0039] In a third aspect, an embodiment of the present application provides a non-transitory computer-readable storage medium storing computer instructions, the computer instructions being used to cause a computer to perform the above-mentioned human-like arm stiffness and motor skill learning method of the robot arm.

[0040] Compared with the prior art, the present application can achieve the following beneficial effects:

[0041] The human-like arm stiffness and motor skill learning method of the robot arm, the computer device and the medium provided in the embodiment of the present application include two stages of demonstration and learning, the demonstration stage collects task trajectory information and human arm stiffness information, the learning stage extracts the trajectory information and the arm stiffness information and establishes corresponding motion trajectory models and adaptive stiffness models, and the control of the robot arm is realized. The motion planning of the task trajectory is completed while learning the human-like stiffness skill, so that the robot arm has the motion skill and the variable stiffness skill at the same time, and can be applied to more working occasions such as human-machine cooperation and auxiliary medical treatment. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 is a flowchart of a human-like arm stiffness and motor skill learning method of a robot arm according to an embodiment of the present application / provided in the specific embodiment of the present application;

[0043] Figure 2 is a structural block diagram of a computer device according to an embodiment of the present application / provided in the specific embodiment of the present application. DETAILED DESCRIPTION

[0044] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. In the following description, the same modules are denoted by the same reference numerals. In the case of the same reference numerals, their names and functions are also the same. Therefore, detailed descriptions thereof will not be repeated.

[0045] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and do not constitute a limitation on the present application.

[0046] In conjunction with Figure 1 the embodiment of the present application provides a human-like arm stiffness and motor skill learning method of a robot arm, which includes:

[0047] S101, wearing an electromyographic signal sensor on the arm of the demonstrator, the electromyographic signal sensor being used to test the arm rigidity information of the demonstrator.

[0048] The skin surface electromyographic signal sensor is pasted on the arm of the demonstrator, and the host computer completes the recording of the arm rigidity by processing the signal of the electromyographic signal sensor. Those skilled in the art should understand that this will not be described in detail.

[0049] S102, installing a force sensor at the end of the mechanical arm and installing a work load, the force sensor being used to measure the weight of the work load to obtain work load weight information.

[0050] The host computer obtains the work load weight information through the force sensor and records it for subsequent adaptive coefficient reference.

[0051] It should be noted that in the present embodiment, S101 and S102 have no sequence restriction and can be performed simultaneously. Alternatively, S102 can be performed first, and those skilled in the art can choose flexibly, which will not be limited.

[0052] S103, the demonstrator drags the mechanical arm from the demonstration starting position to the demonstration target position, and obtains the trajectory information of the mechanical arm and the arm rigidity information in the dragging process, wherein the trajectory information includes time sequence and position information of the end in XYZ three directions.

[0053] During the dragging of the mechanical arm by the demonstrator, the host computer obtains the trajectory information of the mechanical arm and the arm rigidity information in the dragging process, and records the time sequence and the position information of the end in XYZ three directions.

[0054] S104, pre-processing the trajectory information and determining the velocity and acceleration information of each direction at any time according to the position information and the time sequence.

[0055] The trajectory information of the mechanical arm in the demonstration process mainly includes time sequence and position information of the end of the mechanical arm in XYZ three directions. The trajectory information is pre-processed, and the velocity information and acceleration information of each direction at any time are solved according to the position information and the time sequence by using the difference mathematics of discrete time.

[0056] S105, learning the motion trajectory of the mechanical arm by using the DMP algorithm combined with the velocity and acceleration information, and completing the learning of the motion trajectory model in the demonstration stage.

[0057] The mathematical model of DMP is as follows:

[0058] In the formula, represents a constant coefficient; represents a constant coefficient; represents a target position; represents a position at any time; represents a velocity at any time; represents an acceleration at any time; represents a system phase; represents a time scale factor.

[0059] The part before the plus sign in the above formula is a second-order stable convergence system, which ensures that the robot can reach the target point. In order to make the second-order system respond faster and reach the target point, it is necessary to satisfy , which makes the system in a critical damping state. Among them can adjust the running time of the system after learning. When , the motion time of the system is shorter than the time of the demonstration process, and when , the motion time of the system is longer than the time of the demonstration process.

[0060] The part after the plus sign is a nonlinear forcing term. Even if the motion time or the starting point and the target point changes relative to the teaching process, the nonlinear forcing term can ensure that the robot can move with the trajectory shape in the teaching process, has good generalization, and preserves the characteristic information of the teaching trajectory. The nonlinear forcing term formula is as follows:

[0061]

[0062] In the formula, represents a Gaussian kernel function; represents the number of Gaussian kernel functions; represents the weight of each Gaussian kernel function; represents the starting position; represents the center of the Gaussian kernel function; represents the variance of the Gaussian kernel function.

[0063] By determining the number and weight parameters of the Gaussian kernel function, the characteristics of the demonstration trajectory can be learned.

[0064] Among them, the system phase satisfies , which is also called a regular system, is a constant coefficient and satisfies , where is a function of time t. The entire DMP mathematical model is a function of , so the entire system no longer depends on time t. Since the system phase The existence of the DMP makes the mathematical model of multiple DMPs coupled together, driven by a phase at the same time, so that the next step of the stiffness learning model based on the DMP algorithm can be coupled with the motion model, so that the robot has both motion and variable stiffness properties at the same time, realizing the function of having a human arm-like function.

[0065] Through the position information, velocity information and acceleration information of the robot in XYZ three directions at any time obtained in step S104, the constant coefficient in the motion trajectory model is set, and the value of each time can be calculated according to the mathematical model of DMP . .

[0066] Next, the obtained series of values are used to learn the weight of each Gaussian kernel function, so that the value is as close as possible to the series of values obtained by the demonstration trajectory information, that is, the function approximation problem is completed. Wherein the determination problem of the number of Gaussian kernel functions, the more the number is, the higher the fitting accuracy is, but at the same time, the time consumed is more, which can be selected according to the experience combined with the actual problem. By minimizing the error value , the local weighted regression algorithm is adopted to learn the weight parameter in the embodiment, and finally the weight of the i-th Gaussian kernel function is obtained, wherein:

[0067]

[0068] Through the above formula, the estimation of the parameter can be completed, and the determination of the weight parameter in the motion trajectory model is completed, and then only the starting position, target position and running time of the motion trajectory model are set according to the task information of the robot arm, and then the motion trajectory can be generated under the driving of the system phase , thus the learning of the motion trajectory model based on the DMP algorithm is completed.

[0069] S106, using Gaussian kernel function to fit the stiffness information of the arm, and using regular system to drive, adopting local weighted regression algorithm to learn the weight parameter, completing the learning of the stiffness fitting term in the demonstration process.

[0070] Inspired by the DMP-based robot motion trajectory learning algorithm, a DMP-based stiffness learning is proposed, and an adaptive coefficient is proposed, so that the robot has the function of adaptive variable stiffness, and the robot has the performance similar to the human arm, and shows different stiffness according to different loads.

[0071] Inspired by the principle of nonlinear forcing term in the mathematical model of DMP, N Gaussian kernel functions are used to fit the stiffness curve in the teaching process in this patent, and a regular system is used to drive it. It can be expressed as: According to the stiffness information of the arm collected at each moment in the teaching process, the sequence Similarly, by minimizing the error The local weighted regression algorithm is used to learn the weight parameter, and the solution is: In the formula:

[0072]

[0073] Through the above formula, the learning of the weight parameter can be completed, and the formula S obtained can complete the learning of the stiffness fitting term in the teaching process.

[0074] S107, combine the adaptive coefficient with the stiffness fitting term to obtain the adaptive stiffness model in the teaching stage.

[0075] The process in S106 completes the learning of the stiffness fitting term in the teaching process. However, in order to automatically exhibit different stiffnesses when facing different loads like the human arm, the adaptive coefficient concept is proposed in the embodiment of the present application. The adaptive coefficient expression is: In the formula: represents the actual working load weight information; represents the teaching working load weight information.

[0076] Combine the adaptive coefficient with the stiffness fitting term to realize the adaptive stiffness function in the working process, and the formula is as follows: Thus, the adaptive stiffness model in the teaching stage is obtained.

[0077] S108, obtain task information under a new task, the task information includes a starting position, a target position and a motion time, the motion trajectory model outputs a predicted motion trajectory based on the task information, and the adaptive stiffness model outputs predicted stiffness control information based on the task information.

[0078] The task information under the new task, that is, the actual work, can be set according to the work needs, and the learned motion trajectory model and adaptive stiffness model are used to output the predicted motion trajectory and predicted stiffness control information of the robot arm.

[0079] S109, control the robot arm to work based on the predicted motion trajectory and the predicted stiffness control information.

[0080] The human-like arm stiffness and motion skill learning method provided in the embodiment of the present application includes a teaching demonstration stage and a learning stage, the teaching demonstration stage collects task trajectory information and human arm stiffness information, the learning stage extracts the trajectory information and the arm stiffness information and establishes corresponding motion trajectory models and adaptive stiffness models, and the control of the mechanical arm is realized, the motion planning of the task trajectory is completed while the human-like stiffness skill is learned, the mechanical arm has the motion skill and the variable stiffness skill at the same time, and the mechanical arm can be applied to more working occasions such as human-machine cooperation and auxiliary medical treatment.

[0081] Correspondingly, according to the embodiment of the present application, the present application also provides a computer device, a readable storage medium and a computer program product.

[0082] Figure 2 The structural schematic diagram of the computer device 12 provided in the embodiment of the present application is shown. Figure 2 The block diagram of the exemplary computer device 12 suitable for implementing the embodiment of the present application is shown. Figure 2 The computer device 12 shown is merely an example, and should not bring any limitation to the function and use range of the embodiment of the present application.

[0083] As shown in the figure, Figure 2 The computer device 12 is in the form of a general-purpose computing device. The computer device 12 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown in the figure, their connections, and their functions, are merely examples and are not intended to limit the implementations described and / or claimed in this document.

[0084] The components of the computer device 12 can include, but are not limited to, one or more processors or processing units 16, a system memory 28, and a bus 18 that connects the different system components, including the system memory 28 and the processing unit 16.

[0085] The bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor or local bus using any of a variety of bus architectures. By way of example, these architectures include Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

[0086] Computer device 12 typically includes a variety of computer system readable media. Such media can be any available media that is located either internally or externally to computer device 12, including both volatile and nonvolatile media, removable and non-removable media.

[0087] System memory 28 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Computer device 12 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 can be provided for reading from and writing to a non-removable, non-volatile magnetic media (not shown and typically called a "hard drive"). Figure 2 Although not shown, a magnetic disk drive can also be utilized in some embodiments to access and read / write from one or more magnetic disk drives (not shown) that can also be part of computer device 12. As stated above, a disk drive can also be used to read from or write to a non-removable, non-volatile magnetic media (not shown and typically called a "hard drive"). Figure 2 In some embodiments, a magnetic disk drive (not shown) can also be utilized for reading from and writing to a removable, non-volatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive (not shown) can be utilized for reading from or writing to a removable, non-volatile optical disk (such as a CD-ROM, DVD-ROM or other optical media). In these instances, each can be connected to bus 18 by one or more data media interfaces. Storage 28 can include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the application.

[0088] Program / utility 40, having a set (at least one) of program modules 42, can be stored in, for example, memory 28 by way of example, such program modules 42 include an operating system, one or more application programs, other program modules, and program data, each or some combination thereof, which can include implementation of a networking environment. Program modules 42 generally carry out the functions and / or methodologies of embodiments of the application as described herein.

[0089] Computer device 12 can also communicate with one or more external devices 14 such as a keyboard, a pointing device, a display 24, etc.; one or more devices that enable a user to interact with computer device 12; and / or any devices (e.g., network card, modem, etc.) that enable computer device 12 to communicate with one or more other computing devices. Such communication can occur via Input / Output (I / O) interface(s) 22. Still yet, computer device 12 can communicate with one or more networks, such as a local area network (LAN), a general wide area network (WAN), and / or a public network such as the Internet, via network adapter 20. As depicted, network adapter 20 communicates with the other components of computer device 12 via bus 18. It should be appreciated that although not shown, other hardware and / or software modules could be used in conjunction with computer device 12. Examples, include, but are not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0090] Processing unit 16 can execute various function applications and data processing by running programs stored in system memory 28, such as the robot learning method for humanized arm stiffness and motor skills provided by embodiments of the present application.

[0091] The present application also provides a non-transitory computer readable storage medium storing computer instructions, and the computer program is stored on the medium, wherein the program is executed by a processor to implement the robot learning method for humanized arm stiffness and motor skills provided by all embodiments of the present application.

[0092] The computer storage medium of the embodiments of the present application can adopt any combination of one or more computer readable media. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. More specific examples (non-exhaustive list) of computer readable storage medium include: electrical connections having one or more wires, portable computer disks, hard drives, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this document, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0093] A computer readable signal medium can include a propagated data signal with computer executable prograrn code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal can take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium can be any computer readable medium that can be

[0094] Program code embodied on a computer readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire line, optical fiber cable, RF, etc., or any suitable combination of the foregoing. Computer program code for carrying out operations for aspects of the present application can be written in any suitable programming language including object oriented programming languages such as Java, Smalltalk, C++ or the like, conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). These network connections are

[0095] The embodiment of the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the mechanical arm learning method of anthropomorphic arm rigidity and motor skill according to the above.

[0096] It should be understood that the various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the scope of the present disclosure. For example, the steps recited in the present disclosure can be executed in parallel, executed in series, or executed in different orders, and the present disclosure is not limited herein as long as the desired results of the technical solutions of the present disclosure can be achieved.

[0097] The above detailed description does not constitute a limitation on the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present application shall fall within the protection scope of the present application.

Claims

1. A method for learning a robot arm rigidity and motor skill of a humanoid arm, characterized by, The method comprises the following steps: Wearing a myoelectric signal sensor on the arm of a demonstrator, the myoelectric signal sensor being used to test the arm stiffness information of the demonstrator; Installing a force sensor at the end of a mechanical arm and installing a work load, the force sensor being used to measure the weight of the work load to obtain work load weight information; The demonstrator drags the mechanical arm from a demonstration starting position to a demonstration target position, and obtains trajectory information of the mechanical arm and the arm stiffness information during the dragging process, wherein the trajectory information comprises time series and position information of the end in XYZ three directions; Preprocessing the trajectory information and determining the speed and acceleration information of each direction at any time according to the position information and the time series; Learning the motion trajectory of the mechanical arm by using a DMP algorithm combined with the speed and acceleration information, and completing the learning of the motion trajectory model in the demonstration stage; Fitting the time series and the arm stiffness information by using a Gaussian kernel function, driving by using a regular system, learning the weight parameters by using a local weighted regression algorithm, and completing the learning of the stiffness fitting term in the demonstration process; the fitting of the arm stiffness information by using the Gaussian kernel function, the driving by using the regular system, the learning of the weight parameters by using the local weighted regression algorithm, and the completion of the learning of the stiffness fitting term in the demonstration process comprise: The stiffness curve in the teaching process is fitted by using N Gaussian kernel functions, and a regular system is used for driving, which is expressed as: The sequence is obtained according to the stiffness information of the arm collected at each moment in the teaching process, and the weight parameter is learned by using a local weighted regression algorithm by minimizing the error , and the solution is: , wherein: denotes a Gaussian kernel function; denotes the number of Gaussian kernel functions; denotes the weight of each Gaussian kernel function; Completing the determination of the weight parameters by using the above formula, and completing the learning of the stiffness fitting term in the demonstration process; Combining the adaptive coefficient with the stiffness fitting term to obtain an adaptive stiffness model in the demonstration stage; Obtaining task information under a new task, the task information comprising a starting position, a target position and a motion time, the motion trajectory model outputting a predicted motion trajectory based on the task information, and the adaptive stiffness model outputting predicted stiffness control information based on the task information; Controlling the mechanical arm to work based on the predicted motion trajectory and the predicted stiffness control information.

2. The robot learning method of rigidity and motor skill of a humanoid arm according to claim 1, wherein The learning of the motion trajectory of the mechanical arm by using the DMP algorithm combined with the speed and acceleration information, and the completion of the learning of the motion trajectory model in the demonstration stage comprise: The mathematical model of the DMP algorithm is as follows: wherein represents a constant coefficient; represents a constant coefficient; represents a target position; represents a position at an arbitrary time; represents a velocity at an arbitrary time; represents an acceleration at an arbitrary time; represents a system phase; represents a time scale factor; The part before the plus sign in the formula is a second-order stable convergence system, which ensures that the robot arm can reach the target point. In order to make the second-order system respond faster and reach the target point, the following conditions need to be met , which makes the system in a critical damping state, wherein The running time of the system after learning can be adjusted. When , the motion time of the system is shorter than the time of the teaching process, and when , the running time of the system is longer than the time of the teaching process. The part after the plus sign is a nonlinear forcing term, the nonlinear forcing term being used to ensure that the mechanical arm can move in the shape of the trajectory in the demonstration process, and the characteristic information of the motion trajectory in the demonstration process is reserved, and the nonlinear forcing term formula is as follows: In the formula, denotes the starting position; denotes the center of the Gaussian kernel function; denotes the variance of the Gaussian kernel function; by determining the number and weight parameters of the Gaussian kernel function, the learning of the characteristics of the motion trajectory in the teaching process is completed; where the system phase of the regular system satisfies , is a constant coefficient and satisfies , where is a function of time t; The position information, speed and acceleration information of the mechanical arm in XYZ three directions at any time are obtained, the constant coefficient in the mathematical model of the DMP is set, the value of each time is determined according to the mathematical model of the DMP, and the DMP is controlled to realize the motion of the mechanical arm. ;​ Based on the obtained series of values of the Gaussian kernel function, so that the values of the Gaussian kernel function are infinitely close to the series of values obtained through the motion trajectory in the teaching process, wherein the determination problem of the number of the Gaussian kernel functions is that the more the number is, the higher the fitting accuracy is The weight parameter is learned by using a locally weighted regression algorithm, and the weight of the ith Gaussian kernel function is obtained In the formula: The determination of the weight parameters in the model is completed by the above formula, and the learning of the motion trajectory model in the teaching phase is completed, the motion trajectory model being used to generate a motion trajectory under the driving of the system phase according to the start position, the target position and the running time set by the task information of the mechanical arm.

3. The robot learning method of rigidity and motor skill of a humanoid arm according to claim 1, wherein The combination of the adaptive coefficient with the stiffness fitting term to complete the learning of the adaptive stiffness model in the demonstration stage comprises: The adaptive coefficient is combined with the rigidity fitting term to obtain an adaptive rigidity model in a teaching stage, and an expression of the adaptive coefficient is as follows: , wherein: represents actual working load weight information; represents load weight information in a teaching process; and a formula of the adaptive rigidity model is as follows: .

4. A computer device, comprising: The method comprises the following steps: At least one processor; And The memory is in communication connection with the at least one processor; wherein The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the humanized arm stiffness and motion skill learning method of the mechanical arm in any one of claims 1 to 3.

5. A non-transitory computer-readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to execute the humanized arm stiffness and motion skill learning method of the mechanical arm in any one of claims 1 to 3.

Images