Prosthetic control method fusing human-like neuromuscular reflexes and sensory feedback

Abstract

The application provides a prosthesis control method fusing human-like neuromuscular reflex and perception feedback, comprising the following steps: acquiring surface electromyogram signals of amputee's amputation stump; analyzing and processing the surface electromyogram signals to obtain alpha movement instructions; transmitting the alpha movement instructions to a neural paradigm chip; obtaining simulated muscle tension through the neural paradigm chip processing; applying the muscle tension to a prosthesis hand through a pull wire to control finger movement; acquiring perception of the prosthesis hand through a bionic skin module; and transmitting the perception to a multi-modal stimulator, wherein the multi-modal electric stimulator generates corresponding electric stimulation signals and transmits the electric stimulation signals to a surface electric stimulation array electrode to perform real-time transcutaneous electric stimulation on the amputee's sensory area. Thus, the muscle reflex and the perception feedback are fused on the prosthesis hand.

Description

Technical Field

[0001] This invention belongs to the field of prosthetic control technology, specifically a prosthetic control method that integrates human-like neuromuscular reflexes and sensory feedback. Background Technology

[0002] The functional loss caused by amputation has a profound impact on amputees. It not only challenges their daily lives but also affects their career choices and work abilities, limiting certain occupations and potentially requiring them to relearn or adapt to new skills and work environments. However, through the use of prostheses and other assistive devices, amputees can gradually adapt to a new lifestyle, regain confidence, and rebuild their lives. As of the end of 2019, my country had as many as 25 million people with limb disabilities, of whom nearly 3 million were amputees. Amputees experience a significant reduction in their quality of life due to the loss of limb motor function. Therefore, post-amputation assistive rehabilitation is a major social need that urgently needs to be addressed in my country's rehabilitation medicine and disability welfare programs.

[0003] Prosthetic systems are a key technology for assisting amputees in functional rehabilitation and reintegration into society. While various prosthetic hand products are available on the market, many amputees remain unwilling to use them. Existing prosthetic hand control mechanisms differ fundamentally from the human hand's neuromuscular control mechanisms. They lack the neuromuscular reflexes and tactile perception functions of the human hand, making it difficult to adaptively adjust finger stiffness in real time to suit the stiffness of grasping objects, thus greatly limiting the prosthetic hand's operational functions. The human hand, through complex neuromuscular system regulation and tactile feedback, possesses dexterous and compliant motor control capabilities. The dexterity and compliance of human limb motor control allow for dynamic adjustment of muscle force while meeting movement trajectory requirements, thereby achieving the goals of adapting to load, protecting oneself, and resisting interference. The human body's ability to apply force compliantly and dexterously is the result of the synergistic effect of musculoskeletal biomechanics, the tactile perception system, spinal reflexes, and brain regulation. Therefore, prosthetic systems should transcend the limitations of traditional robot trajectory control approaches. By drawing on and mimicking the human body's neuromotor control and tactile feedback mechanisms, and generating and regulating motion in a manner similar to the human body's compliant neuromotor control principles, it is hoped that anthropomorphic compliant motion control of the prosthetic hand can be achieved, thereby leading to a functional leap. Summary of the Invention

[0004] The purpose of this invention is to address the above problems by providing a prosthetic limb control method that integrates human-like neuromuscular reflexes and sensory feedback.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: a prosthetic limb manipulation method integrating human-like neuromuscular reflexes and sensory feedback, comprising:

[0006] The surface electromyography (EMG) signals of the amputee stump are collected in real time using an EMG sensor. The surface EMG signals are then amplified and decoded to obtain alpha movement commands.

[0007] The alpha motion command is sent to the neuromorphic chip, which simulates and analyzes the motion in real time and generates the muscle tension that drives the prosthetic hand.

[0008] By acquiring various physical signals from the prosthetic hand through a bionic skin module installed on the prosthetic hand, different physical signals are converted into different electrical stimulation waveform codes based on the sensory feedback coding method of electrical stimulation finger stimulation.

[0009] The electrical stimulation waveform encoding signal is transmitted to a multimodal electrical stimulator, which generates a corresponding electrical stimulation signal that is transmitted to the surface electrical stimulation array electrode via a sensing feedback information path to provide real-time transcutaneous electrical stimulation to the sensory area of ​​the amputee.

[0010] The neuromorphic chip performs real-time simulation analysis and generates muscle tension to drive the prosthetic hand movements, including:

[0011] The alpha motion command is received through a motor neuron pooling model;

[0012] The motor neuron pool model outputs a spiking nerve signal to the muscle model, and the spiking nerve signal is simultaneously fed back to the intercalated cells, from which the intercalated cells output an inhibitory signal to the motor neuron pool model;

[0013] The muscle model outputs muscle tension signals, as well as muscle length and contraction velocity signals. The muscle tension signals are output to the prosthetic hand to drive its movement. The muscle tension signals are also fed back to the Golgi tendon organ, which outputs Ib primary sensory signals to the motor neuron pool model to regulate the muscle tension output by the muscle model. The muscle length and contraction velocity signals output by the muscle model are fed back to the muscle spindle model, which outputs Ia primary sensory signals to the motor neuron pool model to regulate the muscle tension output by the muscle model.

[0014] In one possible implementation, the muscle tension signal output to the prosthetic hand to drive its movement further includes:

[0015] Real-time acquisition of the length information of the drive cable controlling the prosthetic hand;

[0016] The drive line length signal is then converted into a biomimetic muscle length.

[0017] The bionic muscle length information is then sent to the muscle model.

[0018] In one possible implementation, the sensory feedback encoding method based on electrical stimulation-induced finger sensation converts different physical signals into different electrical stimulation waveform codes, including:

[0019] The bionic skin module converts the collected physical signals into digital signals and then transmits them to the digital signal processor module for information processing.

[0020] The processed signal is converted into different electrical stimulation waveform encoded signals by the multimodal electrical stimulator.

[0021] In one possible implementation, amplifying and decoding the surface electromyography signal to obtain an alpha movement command further includes:

[0022] Obtain gamma motion instructions;

[0023] The output alpha motion command and the output gamma motion command are transmitted together to the muscle spindle model to realize the alpha-gamma co-excitation synergistic regulation characteristics, so as to maintain the sensitivity of the muscle spindle to changes in muscle length.

[0024] In one possible implementation, the motor neuron pool model includes multiple neurons of varying sizes; the muscle model includes two types of muscle fibers: fast-twitch and slow-twitch fibers; the neurons in the motor neuron pool model are used to activate the muscle fibers of the muscle model to generate twitching contractions of different amplitudes.

[0025] In one possible implementation, the surface electrical stimulation array electrodes are disposed in the induced finger sensation area or alternative sensation area of ​​the skin at the amputation stump.

[0026] In some possible implementations, after the bionic skin module on the prosthetic hand acquires various physical signals from the prosthetic hand, it further includes:

[0027] In one possible implementation, the physical signals include pressure signals, shear force signals, temperature signals, humidity signals, and proximity signals.

[0028] If the perceived physical signal is a shear slip signal, the physical signal is converted into a synaptic current analog signal and transmitted to the interneuron. If the threshold potential is reached, a peak potential is generated, and the interneuron outputs a nerve impulse signal to the motor neuron pool model to improve the grip strength of the prosthetic hand.

[0029] If the physical signal is perceived to be a pressure signal or a temperature signal, the signal strength is determined by a signal conversion and processing device. If the threshold of the physical signal exceeds a set value, a nerve impulse signal is sent to the motor neuron pool model to control the release of the prosthetic hand.

[0030] The beneficial effects of this invention are:

[0031] 1. It simulates the neural control and tactile feedback functions of the human hand, effectively integrating human-like neuromuscular reflex control and multimodal tactile perception feedback. This control method possesses both human-like neuromuscular reflex characteristics and multimodal tactile perception feedback functions, enabling anthropomorphic compliant movement control of the prosthetic hand.

[0032] 2. The spinal neuromuscular reflex circuit is simulated in real time using a neuromorphic chip. This chip, built on a customized VLSI architecture, offers significant computational advantages, enabling parallel simulation of the behavior of a large number of spiking neurons without sacrificing speed. In the neuromorphic simulation model, based on the introduction of neuronal differentiation characteristics, fast and slow muscle fiber differentiation is simultaneously realized, fully integrating with the differentiation characteristics of neurons to further recreate the realistic recruitment process of human muscle motor units. Neurons of different sizes can activate muscle fibers to produce twitching contractile forces of varying amplitudes. Furthermore, this simulated spinal neuromuscular reflex circuit includes both muscle spindle feedback loops and Golgi tendon feedback loops.

[0033] 3. The prosthetic hand sensory feedback channel based on electrical stimulation reconstruction can sense not only traditional pressure sensation, but also various physical information such as slip sensation, temperature, and humidity.

[0034] 4. A recurrent inhibition spinal reflex circuit was introduced into the neural chip control system. This recurrent inhibition is achieved through the recurrent synaptic cell, which receives excitatory input from ipsilateral motor neurons and then reciprocates to inhibit their activity, forming a negative feedback loop—recurrent inhibition. Recurrent inhibition plays a crucial role in motor neuron recruitment. When motor neurons cease firing for unpredictable reasons, the excitatory input from the recurrent synaptic cell decreases, weakening the inhibitory effect on ipsilateral motor neurons and making it easier for them to resume firing, thereby improving the overall firing rate of the motor neuron pool model. Recurrent inhibition helps maintain the stability of the firing level in the motor neuron pool model.

[0035] 5. In the neuromorphic simulation model, the important alpha-gamma co-excitation and synergistic regulation characteristics of human neural control are introduced to achieve accurate detection of muscle contraction degree by muscle spindle, which is more in line with the real neuromuscular reflex characteristics of the human body.

[0036] 6. The physical signals collected by the bionic electronic skin are also sent to the neuromorphic chip to generate short spinal reflex regulation. For example, shear slip signals: when there is no slip information, the interneuron has no output. When a shear slip signal is detected, it is simulated as a synaptic current and sent to the interneuron. If the threshold potential is reached, a peak potential is generated, the interneuron fires, and a slip reflex is generated. Then, the nerve impulse is transmitted to the motor neuron pool model, which further regulates muscle contraction and extension in real time, thereby forming a gripping action to avoid slipping in a very short time. For example, temperature signals: when the temperature exceeds the set temperature value, a feedback signal is sent to the neuromorphic chip, which controls the prosthetic hand to release in time without the action of the cerebral cortex, thus achieving short spinal reflex regulation. Attached Figure Description

[0037] Figure 1 A flowchart for a prosthetic hand with neuromuscular reflexes and sensory feedback.

[0038] Figure 2 A block diagram of a control method for a prosthetic hand with neuromuscular reflexes and sensory feedback.

[0039] Figure 3 This is a flowchart of neuromuscular reflex control based on a neuromorphic chip.

[0040] Figure 4 This is a reconstruction diagram of the multimodal sensory feedback channels of a prosthetic hand based on electrical stimulation. Detailed Implementation

[0041] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be described in detail below with reference to the accompanying drawings. The description in this part is only exemplary and explanatory, and should not be used to limit the scope of protection of the present invention in any way.

[0042] refer to Figure 1 This embodiment provides a prosthetic limb manipulation method that integrates human-like neuromuscular reflexes and sensory feedback. Control signals obtained from the amputation stump of an upper limb amputee are transmitted to a neuromorphic chip, which then outputs drive signals to the prosthetic hand. This prosthetic hand is a tendon-driven prosthetic hand. The physical sensory signals obtained by the prosthetic hand are input to a multimodal programmable electrical stimulator. After processing by the multimodal programmable electrical stimulator, the signals are fed back to the sensory cortex of the brain, thereby inducing finger sensation. The physical signals obtained by the prosthetic hand can also be converted and transmitted to the interneuron pool. The interneuron pool judges the intensity of the sensation; if it exceeds a set value, a neural impulse signal is transmitted to the motor neuron pool model to simulate a conditioned reflex.

[0043] refer to Figure 2 The specific method of this embodiment includes:

[0044] S100: The electromyography (EMG) sensor is used to collect the surface EMG signals of the amputee's stump in real time. The surface EMG signals are amplified and decoded to obtain alpha movement commands.

[0045] An electromyography (EMG) sensor is a device that detects and records the electrical activity of muscles. When muscles contract, they generate electrical activity, which can be detected by an EMG sensor. In amputee applications, EMG sensors are typically mounted on the skin surface of the amputation stump to collect real-time data on the electrical activity of the stump muscles. These electrical activity signals, also known as surface electromyography (sEMG), reflect the activity of muscles and nerves in the human body. By decoding these signals, we can understand the amputee's motor intentions, such as the hand or arm movements they intend to make. This information can then be used to drive prostheses to perform corresponding actions.

[0046] Surface electromyography (sEMG) signals are the combined effect of electrical activity in superficial muscles and nerve trunks on the skin surface. The signal is a one-dimensional voltage time-series signal obtained by guiding, amplifying, displaying, and recording the bioelectrical changes of the neuromuscular system during voluntary and involuntary activities via surface electrodes. sEMG signals originate from the bioelectrical activity of spinal cord α-motor neurons under the control of the cerebral motor cortex, and are formed by the temporal and spatial summation of numerous peripheral motor unit potentials.

[0047] Surface electromyography (sEMG) signals can be one-dimensional action potential sequences, which are relatively easy to process from acquired sEMG datasets. They can also be alternating current signals, with amplitudes generally proportional to muscle movement intensity; different movements will produce different sEMGs. Additionally, they can be non-stationary micro-electrical signals with amplitudes ranging from 0 to 1.5 mV, useful signal frequencies from 0 to 500 Hz, and the main energy concentrated in the 20-150 Hz range. These signals typically precede limb movement by 30-150 ms, allowing for advance movement detection. Surface EMG sensors consist of two electrodes attached to the skin at a distance (usually 10 mm) along the muscle fibers. These sensors receive and record the biocurrents generated by muscle contraction on the human body's surface. This information can be used to drive prostheses, enabling them to perform corresponding movements.

[0048] By processing surface electromyography (sEMG) signals, control commands can be derived, which can then be used to drive prosthetic devices. First, sEMG signals are weak electrical currents generated by the contraction of human muscles. These signals are typically in the microvolt (μV) range and therefore require amplification to be processed by subsequent electronic equipment. Then, these amplified signals need to be decoded. The decoding process usually involves signal processing algorithms that extract useful features from the EMG signals, such as amplitude or frequency. These features reflect the activity of muscles and nerves. Finally, these features are converted into alpha motion commands using machine learning algorithms or other control methods. These commands can then be used to control the movement of devices such as prostheses.

[0049] S200: The alpha motion command is sent to the neuromorphic chip, which simulates and analyzes the motion in real time and generates muscle tension to drive the prosthetic hand.

[0050] In one possible implementation, the surface electromyography (EMG) signal is converted into an alpha motion command using a nonlinear Bayesian filter. This alpha motion command is then input into the neuromorphic chip. The Bayesian filter is a nonlinear recursive filter based on Bayesian estimation. The desired filtered signal (the true motion intention) is modeled as a combination of drift and jump processes, while the measured EMG signal is modeled as a stochastic process, with the rate estimated online by computing the complete conditional density of all past measurements. The Bayesian filter incorporates a Poisson jump process, explicitly describing the intention to move rapidly towards the target.

[0051] The neuromorphic chip design mimics the basic architecture of human spinal cord neuromuscular reflexes, placing the computing units (neurons) next to the storage units (synapses connecting neurons). The neuromorphic chip can output realistic muscle tension, which drives the movement of the prosthetic hand.

[0052] refer to Figure 2 The neuromorphic chip performs real-time simulation analysis and generates muscle tension to drive the movements of the prosthetic hand, including:

[0053] S210: Receives alpha motion commands through a motor neuron pooling model.

[0054] The motor neuron pool model is a model in neuromorphic chips that simulates the working mechanism of biological motor neurons. Motor neurons are neurons that activate muscle movement. The motor neuron pool model contains multiple neurons, each of which can be viewed as a processing unit that receives input signals, processes them, and generates an output signal. In the neuromorphic chip, a neuron may correspond to one or more hardware circuits that can process input signals in parallel and generate corresponding output signals. In some possible implementations, neurons are configured with varying sizes to simulate the differentiation characteristics of neurons.

[0055] S220: The spiking neural signals output from the motor neuron pool model are transmitted to the muscle model.

[0056] When a neuron in the motor neuron pool model receives the corresponding alpha movement command, it will generate a nerve impulse and send a pulse nerve signal to the muscle model.

[0057] The muscle model structure of the neuromorphic chip is based on the classic Hill muscle model. The Hill muscle model describes the mechanical behavior of muscles, including how they respond to neural activation, generate force, absorb energy, and control changes in muscle length. In this embodiment, the muscle model contains multiple muscle fibers, which are divided into fast-twitch and slow-twitch fibers according to their contraction speed and capacity. This simulates the differentiation characteristics of fast and slow-twitch muscle fibers, which are fully integrated with the differentiation characteristics of neurons, further reproducing the real recruitment process of human muscle motor units. Neurons of different sizes can activate different fast and slow-twitch muscle fibers to produce twitching contractile forces of varying amplitudes.

[0058] S230: After the pulsatile nerve signal is fed back to the Runshao cell, the Runshao cell outputs an inhibitory signal to the motor neuron pool model.

[0059] In the neuromorphic chip, a spinal reflex loop of recurrent inhibition is introduced. This recurrent inhibition is achieved through the Leysau cells, which receive excitatory input from ipsilateral motor neurons and then send it back to inhibit their activity, forming a negative feedback loop—recurrent inhibition. Recurrent inhibition plays a crucial role in motor neuron recruitment. When motor neurons cease firing for unpredictable reasons, the excitatory input from the Leysau cells decreases, weakening the inhibitory effect on ipsilateral motor neurons and making it easier for them to resume firing, thereby improving the overall firing rate of the motor neuron pool model. Recurrent inhibition helps maintain the stability of the firing level in the motor neuron pool model.

[0060] S240: After the muscle length signal and contraction speed signal output by the muscle model are fed back to the muscle spindle model, the muscle spindle model outputs the Ia primary sensory signal to the motor neuron pool model to adjust the firing level of motor neurons, thereby adjusting the muscle contraction in real time.

[0061] The muscle spindle model here can provide real-time feedback on minute changes in muscle length and contraction speed, enabling sensitive muscle regulation.

[0062] S241: The length of the prosthetic hand drive line is converted and calculated, then converted into the muscle length in the bionic model and sent to the muscle model.

[0063] The joints of the prosthetic hand are controlled by linear motors connected by drive lines. The motors integrate displacement sensors, and the length information of the drive lines can be obtained in real time according to the displacement of the motor shaft. The length signal of the drive lines can be converted into the length signal of the simulated muscles. This length signal and the muscle contraction speed information are first fed back to the muscle model, and then transmitted from the muscle model to the muscle spindle model. The muscle spindle model senses the changes in muscle length in real time and sends neural regulation signals to the motor neuron module.

[0064] S250: After the simulated muscle tension signal is fed back to the Golgi tendon organ, the Golgi tendon organ outputs the Ib primary sensory signal to the motor neuron pool model to adjust the firing level of motor neurons, thereby adjusting muscle tension in real time.

[0065] The main function of the Golgi tendon organ is to regulate muscle tension, maintain the stability of muscle tension, especially for some large tensions, and prevent muscles from being overstretched or damaged.

[0066] In some possible implementations, the control of the neuromorphic chip also includes:

[0067] S260: After amplifying and decoding the surface electromyography signal, the output signal is converted to obtain a gamma motion command. The obtained gamma motion command is transmitted to the muscle spindle model for alpha-gamma co-excitation and coordinated regulation to maintain the sensitivity of the muscle spindle to changes in muscle length and to achieve accurate detection of the degree of muscle contraction by the muscle spindle.

[0068] In the neuromuscular system, alpha and gamma are two different types of motor neurons. Alpha motor neurons directly stimulate skeletal muscle fibers, causing them to contract. They are primarily responsible for fine, conscious movement. Gamma motor neurons stimulate sensory organs in the muscle (called tendon organs and myofibril organs), which sense muscle length and tension. By modulating these sensory organs, gamma neurons can help regulate the muscle's basal tension. In this embodiment, by simultaneously activating alpha and gamma neurons, this synergistic activation ensures that the sensory organs can still correctly sense the muscle's state during muscle contraction.

[0069] S300: Various physical signals from the prosthetic hand are acquired through the bionic skin module set on the prosthetic hand, and different physical signals are converted into different electrical stimulation waveform codes based on the sensory feedback coding method of electrical stimulation finger stimulation.

[0070] refer to Figure 4 The bionic skin in the prosthetic hand is placed at the fingertips. This bionic skin can sense various physical signals, such as pressure, shear force, temperature, humidity, and proximity signals. Based on a multimodal sensory feedback transcutaneous electrical stimulation coding method, it generates different electrical stimulation waveform codes to provide real-time transcutaneous electrical stimulation to the amputee's sensory areas, creating different sensations. This allows the prosthetic hand's motor control to be adjusted according to these sensations, forming a complete closed-loop control system. When the skin senses different stimulation signals, such as touch or temperature changes, the system converts these signals into corresponding electrical stimulation waveform codes according to pre-set coding rules. These electrical stimulation waveform codes can be transmitted transcutaneously to the body, thus stimulating the patient.

[0071] The bionic skin module collects various physical signals (analog signals), converts them into digital signals through an analog-to-digital converter, and then transmits them to a digital signal processor module for information processing. The digital signal processor module converts the physical signals into different electrical stimulation waveform encoded signals.

[0072] In one possible embodiment, the physical signals collected by the bionic electronic skin are also sent to the neuromorphic chip to generate short spinal reflex regulation. For example, shear slip signals: when there is no slip information, the interneuron has no output; when a shear slip signal is detected, it is simulated as a synaptic current and sent to the interneuron. If a threshold potential is reached, a peak potential is generated, the interneuron fires, and a slip reflex is generated. Subsequently, the nerve impulse is transmitted to the motor neuron pool model, which further regulates muscle contraction and extension in real time, thereby forming a gripping action to avoid slipping in a very short time. For example, temperature signals or pressure signals: when the temperature exceeds a set temperature value or the pressure exceeds a set pressure value, a feedback signal is sent to the neuromorphic chip, which controls the prosthetic hand to release in time without the action of the cerebral cortex, thus achieving short spinal reflex regulation.

[0073] S400: The electrical stimulation waveform encoding signal is transmitted to the multimodal electrical stimulator, and the multimodal electrical stimulator generates a corresponding electrical stimulation signal, which is transmitted to the surface electrical stimulation array electrode through the sensing feedback information path to perform real-time transcutaneous electrical stimulation on the sensory area of ​​the amputee.

[0074] Here, the surface electrical stimulation array electrodes are placed in the induced finger sensation area or alternative sensation area of ​​the skin at the amputation stump.

[0075] Among them, the multimodal electrical stimulator can be programmed.

[0076] It should be noted that, in this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0077] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only for the purpose of helping to understand the method and core ideas of the present invention. The above descriptions are only preferred embodiments of the present invention. It should be noted that due to the limitations of textual expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of the present invention.

Claims

1. A method for controlling a prosthetic limb that integrates human-like neuromuscular reflexes and sensory feedback, characterized in that, include: The surface electromyography (EMG) signals of the amputee stump are collected in real time using an EMG sensor. The surface EMG signals are then amplified and decoded to obtain alpha movement commands. The alpha motion command is sent to the neuromorphic chip, which simulates, analyzes, and outputs the muscle tension that drives the prosthetic hand movement in real time. Multiple physical signals are acquired by a bionic skin module set on the prosthetic hand, and different physical signals are converted into different electrical stimulation waveform codes based on the sensory feedback coding method of electrical stimulation finger stimulation. The electrical stimulation waveform encoding signal is transmitted to a multimodal electrical stimulator, which generates a corresponding electrical stimulation signal that is transmitted to the surface electrical stimulation array electrode via a sensing feedback information path to provide real-time transcutaneous electrical stimulation to the sensory area of ​​the amputee. The neuromorphic chip performs real-time simulation analysis and generates muscle tension to drive the prosthetic hand movements, including: The alpha motion command is received through a motor neuron pooling model; The motor neuron pool model outputs a spiking nerve signal to the muscle model, and the spiking nerve signal is simultaneously fed back to the intercalated cells, from which the intercalated cells output an inhibitory signal to the motor neuron pool model; The muscle model outputs muscle tension signals, as well as muscle length and contraction velocity signals. The muscle tension signals are output to the prosthetic hand to drive its movement. Simultaneously, the muscle tension signals are fed back to the Golgi tendon organ, which outputs Ib primary sensory signals to the motor neuron pooling model to regulate the muscle tension output by the muscle model. The muscle length and contraction velocity signals output by the muscle model are fed back to the muscle spindle model, which outputs Ia primary sensory signals to the motor neuron pooling model to regulate the muscle tension output by the muscle model. The muscle model includes two types of muscle fibers: fast-twitch and slow-twitch. The neurons in the motor neuron pool model are used to activate the muscle fibers of the muscle model to generate twitching contraction forces of different amplitudes. After the bionic skin module on the prosthetic hand acquires various physical signals from the prosthetic hand, the method further includes: if the sensed physical signal is a shear slip signal, the physical signal is converted into a synaptic current analog signal by a conversion processing device and transmitted to an interneuron; if a threshold potential is reached, a peak potential is generated, and the interneuron outputs a nerve impulse signal to the motor neuron pool model to improve the grip strength of the prosthetic hand; if the sensed physical signal is a pressure signal or a temperature signal, the signal strength is determined by a signal conversion processing device; if the threshold of the physical signal exceeds a set value, a nerve impulse signal is sent to the motor neuron pool model to control the prosthetic hand to release.

2. The prosthetic limb manipulation method according to claim 1, characterized in that, The method of outputting the muscle tension signal to the prosthetic hand to drive its movement also includes: acquiring the length information of the drive line controlling the prosthetic hand in real time; converting the drive line length signal into a bionic muscle length; and then sending the bionic muscle length information to the muscle model.

3. The prosthetic limb manipulation method according to claim 1, characterized in that, The sensory feedback encoding method based on electrostimulation-induced finger sensation converts different physical signals into different electrostimulation waveform codes, including: the bionic skin module converts the collected physical signals into digital signals and transmits them to the digital signal processor module for information processing; the processed signals are converted into different electrostimulation waveform coded signals by the multimodal electrostimulator.

4. The prosthetic limb manipulation method according to claim 1, characterized in that, The surface electromyography signal is amplified and decoded to obtain an alpha movement command, and the method further includes obtaining a gamma movement command; the output gamma movement command is transmitted to the muscle spindle model to realize the alpha-gamma co-excitation synergistic regulation characteristics, so as to maintain the sensitivity of the muscle spindle to changes in muscle length.

5. The prosthetic limb manipulation method according to claim 1, characterized in that, The motor neuron pool model includes multiple neurons of varying sizes.

6. The prosthetic limb manipulation method according to claim 1, characterized in that, The surface electrostimulation array electrodes are positioned in the induced finger sensation area or alternative sensation area of ​​the skin at the amputation stump.

7. The prosthetic limb manipulation method according to any one of claims 1-6, characterized in that, The physical signals include pressure signals, shear slip signals, and temperature signals.

Images