Tactile nociceptor circuit based on memristor and application of tactile nociceptor circuit

Abstract

The invention belongs to the technical field of artificial intelligence, and particularly discloses a tactile nociceptor circuit based on a memristor and application of the tactile nociceptor circuit. The circuit comprises a nociceptor module which is used for simulating a dynamic balance process of a membrane potential of a biological nociceptor, detecting external stimulation and generating a corresponding pulse signal by using a nonlinear resistance state change characteristic; the external stimulation comprises physical, temperature and chemical stimulation; the pain grading module adopts a logic circuit and is used for integrating the pulse signals output by the nociceptor module, realizing pain grading and outputting pain signals representing slight pain and severe pain; and the amygdaloid nucleus pathway module is constructed based on a BEL model and is used for dynamically generating a titer signal associated with the pain signal. The hierarchical information processing mechanism of the biological nervous system can be reproduced through system-level circuit design, and an extensible hardware basis is provided for emotional learning and memory enhancement research of a subsequent brain-like intelligent system.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of artificial intelligence, and more particularly relates to a tactile nociceptor circuit based on a memristor and application thereof. BACKGROUND

[0002] As a technology system simulating human intelligence, artificial intelligence has reached or surpassed human level in image recognition, target detection, medical image analysis and other fields. However, due to the storage-computation separation mechanism of traditional computing architecture, artificial intelligence still cannot simulate the efficient information processing capability of the biological nervous system, which affects its further evolution to high-order cognitive and adaptive learning capability.

[0003] The ability to perceive external harm is a key starting point and core driving factor for biological learning and memory. When a biological body encounters external harm, the harm stimulus will trigger a series of complex neural responses, prompting the brain to quickly adjust the behavior pattern and strengthen the memory. Similarly, if a brain-like intelligent system can simulate this external harm perception ability, it can build a more realistic and efficient learning and memory model based on it. Through accurate capture, processing and storage of harm signals, brain-like intelligence can achieve rapid response to dangerous environments. In the harm perception scene, touch as a direct physical contact perception modality can capture potential threats first and trigger a millisecond-level rapid avoidance reaction. Researching to preferentially build a touch perception hardware to simulate the core function of real-time interactive defense of a biological body significantly improves the survival ability and task execution efficiency of an intelligent agent in a complex environment, and provides an important reference for the research of brain-like intelligent systems.

[0004] Current research on harm perception based on memristors mainly focuses on device material innovation, such as using Ag / TiOx / Pt diffusion type memristors to simulate threshold accumulation effect, two-dimensional MoS memristors to improve response speed, and FK-800 copolymer memristors to realize pain sensation bionics. However, these researches focus on the functional optimization of single devices, lack of system-level circuit design for the hierarchical information processing mechanism of biological nervous systems, and device innovation can only achieve local property bionics, but lack of adaptive circuit architecture support, making it difficult to reproduce the complex hierarchical mechanism of biological harm perception.

[0005] Therefore, how to solve the bionics problem of the complex hierarchical mechanism of the biological nervous system is an urgent research need. SUMMARY

[0006] In view of the defects of the prior art, the purpose of the present application is to provide a tactile nociceptor circuit based on a memristor and application thereof, which can reproduce the hierarchical information processing mechanism of the biological nervous system through system-level circuit design, and provide an extensible hardware foundation for the emotional learning and memory reinforcement research of subsequent brain-like intelligent systems.

[0007] To achieve the above object, in a first aspect, the application provides a tactile nociceptor circuit based on a memristor, comprising:

[0008] a nociceptor module for simulating the dynamic balance process of the membrane potential of a biological nociceptor, detecting external stimuli and generating corresponding pulse signals by using a nonlinear resistance state change characteristic; the external stimuli include physical, temperature and chemical stimuli;

[0009] a pain grading module for integrating the pulse signals output by the nociceptor module, realizing pain grading and outputting pain signals representing mild pain and severe pain by using a logic circuit;

[0010] an amygdala pathway module for dynamically generating a valence signal associated with the pain signal based on a BEL model.

[0011] As a further preferred, the nociceptor module includes three different types of nociceptor circuits of touch, temperature and chemistry, each of which includes a memristor M1, an OR gate OR1, NMOS tubes N1-N5 and PMOS tubes P1-P5.

[0012] In each nociceptor circuit, the anode of the memristor M1 is connected to the drain of the PMOS tube P1 and the source of the NMOS tube N1, the cathode of the memristor M1 is connected to the gate of the PMOS tube P2, the gate of the NMOS tube N4 and the gate of the NMOS tube N5, and is connected to a fixed resistor R1; the PMOS tube P1 and the NMOS tube N1 are connected in common gate and are connected to the output end of the OR gate OR1, the drain of the NMOS tube N1 is connected to the source of the NMOS tube N2, the source of the PMOS tube P1 and the gate of the NMOS tube N2 are connected in common to form the input end of the nociceptor circuit, and the drain of the NMOS tube N2 is connected to a-1.5V power supply; the PMOS tube P2, the NMOS tube N3 and the PMOS tube P3 form a comparator COMP1, the NMOS tube N4 and the PMOS tube P4 form a comparator COMP2, the NMOS tube N5 and the PMOS tube P5 form a comparator COMP3, the output ends of the comparator COMP1 and the comparator COMP2 are connected to the two input ends of the OR gate OR1, and the output end of the comparator COMP3 forms the output end of the nociceptor circuit.

[0013] As a further preferred, each nociceptor circuit generates an action potential under the continuous action of the corresponding external stimulus, and has four processes of depolarization, repolarization, hyperpolarization and reset in each action generation process, for simulating the dynamic balance process of the membrane potential of a biological nociceptor.

[0014] As a further preferred, each pulse generated by each nociceptor has four stages:

[0015] Na+ ions permeate into the cell, making the cell membrane enter the depolarization stage, and the cell membrane potential starts to rise, generating a peak voltage;

[0016] When the membrane potential reaches the peak, the Na+ channel closes, and K+ ions start to flow out of the cell, causing the cell membrane to enter the repolarization, and the membrane potential starts to drop;

[0017] Due to the permeability of K+ channels, the cell membrane potential drops below the resting potential, i.e. the hyperpolarization stage;

[0018] In the absence of external stimulus input or low stimulus intensity, the membrane potential returns to the resting state again.

[0019] As a further preferred, the pain grading module comprises a NOT gate, a NAND gate and AND gates AND1-AND3;

[0020] Wherein, one input terminal of the AND gate AND1 receives the pulse signal output by the nociceptor module under external physical stimulation, the other input terminal of the AND gate AND1 receives the pulse signal output by the nociceptor module under external temperature stimulation, and the two input terminals of the AND gate AND1 are connected to the two input terminals of the NAND gate; the input terminal of the NOT gate and one input terminal of the AND gate AND3 both receive the pulse signal output by the nociceptor module under external chemical stimulation, the output terminal of the NOT gate is connected to one input terminal of the AND gate AND2, the other input terminal of the AND gate AND2 is connected to the output terminal of the AND gate AND1, the other input terminal of the AND gate AND3 is connected to the output terminal of the NAND gate, and the output terminals of the AND gate AND2 and the AND gate AND3 constitute the two output terminals of the pain grading module.

[0021] As a further preferred, the amygdala pathway module comprises two amygdala pathway circuits, each of which comprises a memristor M2-M4 and an operational amplifier OP1 and OP2;

[0022] In each amygdala pathway circuit, the memristor M2 is connected in series with the memristor M3, the positive electrode of the memristor M2 is the input terminal of the amygdala pathway circuit, and the negative electrode of the memristor M2 is connected to a fixed resistor R9 and a capacitor C1; the negative electrode of the memristor M3 is connected to the inverting input terminal of the operational amplifier OP1, the non-inverting input terminal of the operational amplifier OP1 is grounded, and the output terminal of the operational amplifier OP1 is connected to the non-inverting input terminal of the operational amplifier OP2 through a negative feedback resistor R 10 The output terminal of the operational amplifier OP1 is connected to the inverting input terminal of the operational amplifier OP2, and the non-inverting input terminal of the operational amplifier OP2 is grounded. 11 The output terminal of the operational amplifier OP2 is connected to the non-inverting input terminal of the operational amplifier OP1 through a negative feedback resistor R 12The positive electrode of the memristor M4 is connected to the output end of the operational amplifier OP2, and the negative electrode of the memristor M4 is connected to a fixed resistor R 13 , and the output end of the almond kernel path circuit.

[0023] As a further optimization, the value of the valence signal depends on the pain signal and the memristive value of the memristors M2 and M3.

[0024] In a second aspect, the application provides an application of the above-mentioned memristor-based tactile nociceptor circuit, which is applied to the fields of prosthetics, robot safety interaction and neuromorphic chips.

[0025] As a further optimization, the neuromorphic chip is deployed on a wearable device.

[0026] As a further optimization, the wearable device includes an athlete's smart protective gear and a firefighter's uniform.

[0027] When deployed on the athlete's smart protective gear, it is used to monitor the athlete's muscle strain risk in real time.

[0028] When deployed on the firefighter's uniform, it is used to monitor the degree of thermal injury accumulation.

[0029] Compared with existing technologies, the application has the following advantages:

[0030] (1) Bionic and systematic improvement: Breakthrough the local optimization limitation of single device in existing technologies, reproduce the hierarchical information processing mechanism of biological nervous system through system-level circuit design, and realize the deep integration of nociception and emotion generation.

[0031] (2) Enhanced dynamic adaptability: Based on the threshold dynamic adjustment function of the memristor, the dynamic adaptation characteristics of biological nociceptors can be simulated, which significantly improves the response accuracy and robustness of the circuit to complex external stimuli.

[0032] (3) Energy efficiency and integration optimization: Using the non-volatile storage characteristics of the memristor and the low-power advantage of the spiking neural network, the traditional ADC module is omitted, which significantly reduces the system power consumption and reduces the hardware area occupation.

[0033] (4) Multimodal injury grading capability: Through the integration of physical, temperature and chemical stimulation signals by logic circuit, the injury grading (such as severe pain, mild pain) and the association of emotional valence signal (such as sadness, joy) are realized, which simulates the physiological linkage mechanism of biological "pain-emotion".

[0034] (5) Biological compatibility expansion: The circuit design is compatible with memristor devices made of different materials, which has strong universality and provides an expandable hardware foundation for subsequent emotional learning and memory enhancement research of brain-like intelligent systems. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a circuit schematic diagram of a haptic nociceptor circuit based on a memristor provided by an embodiment of the present application;

[0036] Figure 2 is a circuit schematic diagram of one of the nociceptor circuits provided by an embodiment of the present application;

[0037] Figure 3 is a depolarization, repolarization, and hyperpolarization process diagram of a membrane voltage generating action potential provided by an embodiment of the present application;

[0038] Figure 4 is a reset process diagram of a membrane voltage generating action potential provided by an embodiment of the present application;

[0039] Figure 5 is a circuit schematic diagram of one of the amygdala pathway circuits provided by an embodiment of the present application;

[0040] Figure 6 is a circuit schematic diagram of a pain grading module provided by an embodiment of the present application;

[0041] Figure 7 is a circuit simulation result diagram of perceiving a severe pain injury provided by an embodiment of the present application;

[0042] Figure 8 is a circuit simulation result diagram of perceiving a slight pain injury provided by an embodiment of the present application;

[0043] FIG. 9(a) is a circuit simulation result diagram of generating a negative emotion provided by an embodiment of the present application;

[0044] FIG. 9(b) is a circuit simulation result diagram of generating a positive emotion provided by an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will be combined with the accompanying drawings for the embodiments of the present application to clearly and completely describe the technical solutions of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person skilled in the art without creative work fall within the scope of protection of the present application.

[0046] It should be noted that the memristor refers to a memory function resistor device, which can change its resistance value after applying voltage, and "remember" its state after the power is turned off, thereby realizing non-volatile storage. The resistance value of the memristor will change to the corresponding resistance state with the external operation and the previous state of itself, for example, in the high resistance state, after being stimulated by a strong positive electric stimulus, it will suddenly change to a low resistance state. Such resistance transition characteristics can be well explained by the mechanism of nanoscale conductive filament formation and rupture. When the external electrical stimulus is strong enough, ions move to form a conductive filament, and the resistance of the memristor changes suddenly. When the external electrical stimulus is lower than the critical value, the ions do not move, and no conductive filament is formed, so the resistance of the memristor will not change suddenly.

[0047] To reproduce the complex hierarchical mechanism of biological nociception, the present application provides a memristor-based tactile nociceptor circuit, which includes a nociceptor module, a pain grading module and an amygdala pathway module.

[0048] Among them, the nociceptor module provided in the embodiment is used to simulate the dynamic balance process of the membrane potential of the biological nociceptor by using the nonlinear resistance state change characteristic, detect external stimuli (including physical, temperature and chemical stimuli) and generate corresponding pulse signals.

[0049] In the embodiment, the circuit design of the nociceptor module is as follows:

[0050] Nociceptor is a kind of sensory neuron that adaptively processes external data in time domain. This means that nociceptor has a dynamic internal "state", which is very consistent with the nonlinear characteristics of the memristor. In essence, both biological nociceptors and memristors can process information in a changing and environment-dependent manner. Based on this similarity, the key function of biological nociceptors, amplifying external stimuli according to certain criteria, can be effectively realized by the memristor. In the memristor nociceptor, the voltage signal applied to the memristor is used to simulate external stimuli. When the voltage amplitude exceeds the threshold voltage of the memristor, the memristor changes to a low resistance state, and a current pulse is detected at the output end, which corresponds to the pain perception of harmful stimuli. Studies have shown that when the intensity of harmful stimuli exceeds the threshold of nociceptors, the firing rate of nociceptors will increase accordingly with the increase of the intensity of the stimulus, thereby reflecting the severity of harmful stimuli. Conversely, if the input voltage is not enough to make the memristor change to a low resistance state, the memristor remains in its initial high resistance state, and at this time no current passes through the output end, indicating that the external stimulus does not reach a harmful degree.

[0051] In the traditional neuron circuit, the capacitor is usually used as the physical carrier of the cell membrane potential, and the dynamic change mechanism of the biological membrane potential is simulated by the charging and discharging process between the two poles of the capacitor. In this embodiment, a memristor is introduced to replace the capacitor, and its unique physical characteristics are that the directional migration of internal carriers will produce nonlinear resistance state changes when a forward / reverse bias voltage is applied. This dynamic regulation characteristic can more truly reproduce the dynamic balance process of the membrane potential.

[0052] The nociceptor module provided in this embodiment includes three different types of nociceptor circuits, namely physical, temperature and chemical nociceptor circuits, and each nociceptor circuit includes a memristor M1, an OR gate OR1, NMOS tubes N1-N5 and PMOS tubes P1-P5.

[0053] As shown in Figure 2 , the device connection relationship is as follows: V In is a tactile input signal, V Tac is a tactile output signal of the circuit, the positive electrode of the memristor M1 is connected with the drain electrode of the PMOS tube P1 and the source electrode of the NMOS tube N1, the negative electrode of the memristor M1 is connected with the gate electrode of the PMOS tube P2, the gate electrode of the NMOS tube N4 and the gate electrode of the NMOS tube N5, and is connected with a fixed resistor R1=1kΩ; the common gate of the PMOS tube P1 and the NMOS tube N1 is connected with the output end of the OR gate OR1, the drain electrode of the NMOS tube N1 is connected with the source electrode of the NMOS tube N2, the source electrode of the PMOS tube P1 and the gate electrode of the NMOS tube N2 are commonly connected to form the input end of the nociceptor circuit, and the drain electrode of the NMOS tube N2 is connected with a-1.5V power supply; the PMOS tube P2, the NMOS tube N3 and the PMOS tube P3 constitute a comparator COMP1, the NMOS tube N4 and the PMOS tube P4 constitute a comparator COMP2, the NMOS tube N5 and the PMOS tube P5 constitute a comparator COMP3, the output ends of the comparator COMP1 and the comparator COMP2 are correspondingly connected with the two input ends of the OR gate OR1, and the output end of the comparator COMP3 constitutes the output end of the nociceptor circuit. 2-8 =10kΩ.

[0054] Generally speaking, in some studies on the perception of electrical stimulation on the skin surface, it is shown that the perception of electrical stimulation on the skin depends on sufficient voltage or current intensity. Under low voltage or current, most individuals have difficulty in producing clear tactile response due to high skin impedance and unsatisfied nerve activation threshold. Taking the skin of human fingers as an example, when the current amplitude is 0.43mA, the human body has a slight electric stimulation feeling when the impedance value is 3.5kΩ; when the current increases to 1.10mA, a larger pressure and vibration feeling is produced; and when the current reaches 1.45mA, the electric stimulation makes the human body feel pain.

[0055] Take the input signal VIn=1.51V as an example to analyze the working principle of the tactile nociceptor circuit:

[0056] V GS-P1 =V3-V In =V3-1.51<V TH-P1 (3.1)

[0057] V GS-N1 =V3-V1=V3-V In =V3-1.51<V TH-N1 (3.2)

[0058] At this time P1 is on, N1 is off, VIn is applied to the memristor M1, which exceeds the positive threshold voltage of the memristor, and the resistance gradually decreases from the high resistance state (about 9kΩ). The negative voltage V2 of M1 represents the cell membrane potential, and its expression is:

[0059]

[0060] Where M(t) can be seen in Section 2.3 formula (2.12), when M1 gradually decreases, V2(0)=0.151V gradually increases. P2, N3, P3 constitute a comparator COMP1:

[0061]

[0062] Where -0.1V is the threshold voltage of the cell membrane potential, N4, P4 constitute a comparator COMP2:

[0063]

[0064] N5, P5 constitute a comparator COMP3:

[0065]

[0066] Only when V2 > 0.2V, N5 is on, at this time, comparator COMP3 outputs high level VTac = 1V. When V2 continues to increase to 0.4V, comparator COMP2 outputs high level, thus or gate OR1 outputs high level. At this time, P1 is off, N1, N2 are on, the drain voltage of N2 -1.5V is applied to the memristor M1, V2 changes from positive potential to negative potential, and exceeds the negative threshold voltage of the memristor, so that the memristor gradually increases from the low resistance state. As can be seen from formula (3.3), when V1 is negative, M1 gradually increases, and V2 also gradually increases. Before V2 increases to the threshold voltage of the membrane potential -0.1V, comparator COMP1 outputs high level, and or gate OR1 outputs high level, so that N1 continues to be on, and -1.5V is always applied to the memristor M1. Until V2 increases to exceed the threshold voltage of the membrane potential -0.1V, comparator COMP1 outputs low level, comparator COMP2 also outputs low level, thus or gate OR1 outputs low level, so that P1 is on again, N1 is off again, VIn is applied to the memristor M1 again, V2 changes from negative potential to positive potential again, comparator COMP3 outputs low level VTac = 0V, and the cycle is repeated. When there is no input from the outside world, i.e. VIn = 0V, N1 and P1 are both off, and no output signal is generated. This process is equivalent to the nociceptor returning to the resting state, so as to be ready for the next stimulation.

[0067] One of the biggest features of the nociceptor is "threshold response", when the external harmful stimulus does not reach the threshold, no action potential is generated, which indicates that the external stimulus received at this time is harmless. This feature can also be simulated by the memristor. By adjusting the positive threshold voltage of the memristor, when the external stimulus input cannot exceed the threshold voltage of the memristor, the memristor maintains the high resistance state and the circuit has no response. See Figure 3 and Figure 4 .

[0068] The bit V2 gradually rises. When V2 > 0.2V, N3 is on, at this time, high level VTac = 1V is output, and a pulse signal is generated. Until V2 increases to 0.4V, the drain voltage of N2 -1.5V acts on the memristor M1, the membrane potential V2 changes from positive to negative, and the memristor value increases. When V2 increases to -0.1V, which exceeds the threshold voltage of the membrane, VIn is applied to the memristor M1 again, V4 changes from negative potential to positive potential again, and the cycle is repeated.

[0069] It can be observed that under the continuous action, the membrane voltage V2 changes regularly, i.e. the nociceptor generates an action potential, and there are four processes of depolarization, repolarization, hyperpolarization and reset in each process of generating an action potential. Figure 3 and Figure 4The four processes of membrane voltage generating action potential were simulated, and the input information was encoded as pulse discharge rate. Each pulse generated by the nociceptor has four stages:

[0070] (1) Na+ ions penetrate into the cell, causing the cell membrane to enter the depolarization phase, and the cell membrane potential begins to rise, generating a peak voltage.

[0071] (2) When the membrane potential reaches its peak, the Na+ channel closes and K+ ions begin to flow out of the cell, causing the cell membrane to enter repolarization and the membrane potential to begin to decrease.

[0072] (3) Due to the permeability of K+ channels, the cell membrane potential drops below the resting potential, i.e., the hyperpolarization phase;

[0073] (4) When there is no external stimulus input or the stimulus intensity is very low, the membrane potential returns to the resting state.

[0074] Subsequent tests with tactile input signals of varying intensities (VIn = 3.85V and 5.08V) revealed that the same action potentials were generated, albeit at different frequencies. Figure 4 As can be seen, the pulse discharge rate decreases as the input signal strength decreases. The highest frequency, i.e., the highest pulse discharge rate, is observed at VIn = 5.08V, while the lowest frequency, i.e., the lowest peak discharge rate, is observed at VIn = 1.51V. The pulse discharge rate also refers to the number of pulses encoded into an external stimulus within a time window; therefore, the pulses output by this nociceptor module carry input information. Furthermore, observing the changes in memristor value under three different signal intensities reveals that the lower the input signal strength, the greater the rate of change in memristor value M1, indicating a greater amplification of the dangerous external stimulus. This response simulates the "sensitization" characteristic common in biological receptors; that is, when the stimulus is intense, the nociceptor increases pain sensitivity and leads to tissue damage, at which point even a light touch on the skin causes severe pain.

[0075] The amygdala pathway module provided in this embodiment is constructed based on the BEL model and is used to dynamically generate valence signals associated with the pain signals.

[0076] In this embodiment, the circuit design of the amygdala pathway module is as follows:

[0077] In the human brain, the amygdala, as the core hub of the limbic system, performs a dual function: evaluating emotional valence and storing injury memories. It dynamically modulates negative emotional responses such as fear and aversion by integrating impulse signals from pre-nociceptors and forms injury-behavior associations based on synaptic plasticity. Traditional emotional circuits often employ fixed-weight linear synaptic models, which struggle to simulate the amygdala's dynamic weighting and tuning characteristics of spatiotemporal impulse patterns.

[0078] The embodiment is based on the BEL model to construct the amygdala pathway module, which includes two amygdala pathway circuits, each of which includes a memristor M2-M4 and an operational amplifier OP1 and OP2, and the specific circuit design is as follows Figure 5 .

[0079] In each amygdala pathway circuit, VNoc is a nociception signal, VEmo (Emotion) is a corresponding valence signal generated, the memristor M2 is connected in series with the memristor M3, the positive electrode of the memristor M2 is the input end of the amygdala pathway circuit, and the negative electrode of the memristor M2 is connected with a fixed resistor R9 = 5kΩ and a capacitor C1 = 800μF; the negative electrode of the memristor M3 is connected with the inverting input end of the operational amplifier OP1, the non-inverting input end of the operational amplifier OP1 is grounded, the output end of the operational amplifier OP1 is connected with the inverting input end thereof through a negative feedback resistor R 10 = 1kΩ, and the output end of the operational amplifier OP1 is connected with the inverting input end of the operational amplifier OP2 through a fixed resistor R 11 = 1kΩ, the non-inverting input end of the operational amplifier OP2 is grounded, and the output end of the operational amplifier OP2 is connected with the inverting input end thereof through a negative feedback resistor R 12 = 1kΩ; the positive electrode of the memristor M4 is connected with the output end of the operational amplifier OP2, and the negative electrode of the memristor M4 is connected with a fixed resistor R 13 = 5kΩ and is the output end of the amygdala pathway circuit.

[0080] The primary somatosensory cortex receives and processes the impulse signals from the nociceptor output to form emotion generation signals and transmits them to the amygdala, which functions like a synapse. When receiving signals of a certain intensity, the memristive value of the memristor M2 changes accordingly, and the smaller the memristive value is, the greater V7 is, and the fixed resistor R 10 functions to convert the current signal into a voltage signal and output it to the subsequent amygdala module. For the amygdala module, the memristor M3 and the capacitor C1 are connected in parallel, and the capacitor functions to filter the final output signal. Under the influence of the emotion generation signal V7, the memristive value of M3 changes from high to low, that is, the corresponding synaptic weight changes from low to high, thereby generating an emotion signal V8.

[0081]

[0082] OP3, M3 and R 11The function of proportional operation is realized by Kirchhoff's voltage law and operational amplifier knowledge, and the expressions of V7-9 and VEmo can be obtained as shown in formulas (3.7), (3.8), (3.9), (3.10). It can be seen that the value of the potency signal VEmo is related to the input signal VNoc and the memristor value of the memristor M2-3, which reflects the learning changes caused by pain in the brain, which can trigger and affect the expression of emotions.

[0083] The overall circuit design provided by the embodiment is as follows:

[0084] The tactile nociceptors in the human body can detect various external environmental information. The previously designed receptors are mainly used to perceive the intensity of external physical stimuli such as pressure and stinging. Some tactile nociceptors are sensitive to temperature changes and can detect whether the environmental temperature is too high or too low, thereby avoiding damage to the body caused by heat or cold. In addition, when certain special chemicals come into contact with the skin, they can activate specific receptors and affect the excitability of the brain to achieve the effect of reducing pain. For example, menthol can activate the TRPM8 channel, which is an ion channel activated by cooling and mint stimulation. The activation of the TRPM8 channel can produce a cooling sensation.

[0085] The judgment of pain intensity is a complex process, which is not determined by the activation of a single receptor, but is the result of the combined activation of multiple different types of receptors. There are various receptors in the skin and body tissues, each responsible for detecting different types of stimuli such as mechanical pressure, temperature changes, and chemicals. When external stimuli act on the body, these receptors will be activated to different degrees according to the nature and intensity of the stimulus. Figure 7 and Figure 8 .

[0086] In order to distinguish different degrees of external harm, the embodiment designs a pain grading module, which uses a logic circuit to integrate the pulse signals output by the nociceptor module and realizes pain grading to output pain signals representing mild pain and severe pain.

[0087] As shown in Figure 4 , the logic circuit includes NOT gates, NAND gates, and AND gates AND1-AND3, which can receive and process tactile pulse signals VTac1-3 generated by the encoding of three different types of tactile nociceptors. VTac1 is the signal output by the external physical stimulus receptor, VTac2 is the signal output by the temperature receptor, and VTac3 is the signal output by the chemical stimulus channel receptor.

[0088] In the embodiment, the connection relationship of the devices in the logic circuit is as follows: one input end of the AND gate AND1 receives the pulse signal output by the nociceptor module under external physical stimulation, the other input end of the AND gate AND1 receives the pulse signal output by the nociceptor module under external temperature stimulation, and the two input ends of the AND gate AND1 are connected to the two input ends of the NAND gate NAND; the input end of the NOT gate NOT and one input end of the AND gate AND3 both receive the pulse signal output by the nociceptor module under external chemical stimulation, the output end of the NOT gate NOT is connected to one input end of the AND gate AND2, the other input end of the AND gate AND2 is connected to the output end of the AND gate AND1, the other input end of the AND gate AND3 is connected to the output end of the NAND gate NAND, and the output end of the AND gate AND2 and the output end of the AND gate AND3 constitute the two output ends of the circuit.

[0089] According to the analysis of the above circuit, the value of VTac only has two cases of high level 1V and low level 0V, and the output high level indicates that the receptor receives the stimulation signal and is activated, so according to the different input conditions, the truth table of VTac1, VTac2, VTac3, AND2 and AND3 is established, and the level response is related to the pain intensity, as shown in Table 3.1.

[0090] Table 3.1 Truth table of nociceptive grading logic circuit

[0091]

[0092]

[0093] As can be seen from Table 3.1, only when VTac1 and VTac2 output high level at the same time, and VTac3 outputs low level, AND2 outputs high level, and in other cases, it outputs low level, at this time, the external physical stimulation receptor and the temperature receptor are activated, and the chemical stimulation receptor is not activated; only when VTac1 and VTac2 output low level at the same time, and VTac3 outputs high level, AND3 outputs high level, and in other cases, it outputs low level, at this time, the external physical stimulation receptor and the temperature receptor are not activated, and the chemical stimulation receptor is activated alone. In the embodiment, the output high level of AND2 corresponds to the recognition of strong nociception, and the human body feels severe pain. The output high level of AND3 corresponds to the recognition of weak nociception, and the human body feels slight pain due to the effect of chemicals, so the circuit well realizes the pain grading function participated by multiple nociceptors. Figure 7 and Figure 8 .

[0094] Based on the biomimetic function construction of the prelude module, this section designs the pain grading and emotion generation circuit through multi-level memristor circuit collaborative design, aiming at the hierarchical characteristics of biological injury signal transduction, innovatively adopts the three-level architecture of "nociceptive encoding-pain grading-emotion mapping", and designs the pain grading and emotion generation circuit. The specific circuit schematic diagram is as shown in Figure 1 .

[0095] The output signal VNoc1 of the circuit represents the severe pain signal, at this time the corresponding generated valence signal VEmo1 is sadness and anger, VNoc2 represents the mild pain signal, at this time the corresponding generated valence signal VEmo2 is pleasure. It can be known from formula (3.10) that the valence signal VEmo has clear positive and negative properties, and its positive can intuitively reflect the level of emotional pleasure, and this level ranges from one end of extreme sadness to the other end of extreme joy. When VEmo>0, the corresponding degree of pleasure is higher, which is a positive emotion; if VEmo<0, the degree of pleasure decreases, which is a negative emotion, and this binary nature constitutes the basic framework of emotional evaluation.

[0096] The function verification of tactile nociception is the core link to ensure the effectiveness and reliability of the bionic circuit design. Based on the tactile nociceptive signal processing mechanism of biological nervous system, this embodiment carries out function verification from three aspects of action potential encoding of nociceptor, pain intensity grading and emotion generation. The simulation verification results are as shown in Figures 3-4 , 7-9.

[0097] (1) Pain intensity grading: VIn1 is the signal received by the physical stimulus receptor, VIn2 is the signal received by the temperature receptor, and VIn3 is the signal received by the chemical stimulus receptor. When the pressure gradient signal VIn1>1.5V, the temperature change signal VIn2>1.2V and the chemical signal VIn3<0.9V, the system determines that the external stimulus causes severe pain; when the pressure gradient signal VIn1<1.5V, the temperature change signal VIn2<1.2V and the chemical signal VIn3>0.9V, the system determines that the external stimulus causes mild pain. In order to truly simulate these two different injury states, this paper sets two different groups of injury intensity input signals. In the first group of signals, the values of VIn1-3 are set to 1.52V, 1.22V and 0.88V respectively, which can activate the pressure receptor and the temperature receptor, and inhibit the chemical receptor; and the values in the second group of signals are 1.48V, 1.18V and 0.92V respectively, which correspond to the inhibition of the pressure receptor and the temperature receptor, and the activation of the chemical stimulus receptor.

[0098] Through the simulation of the experiment, the Figure 7The results are shown. During the first set of experiments, the first set of signals is input. During this period, the baroceptor and thermoreceptor generate pulse signals VTac1 and VTac2, indicating that both perceive harmful stimulation to be activated, because the signal intensity they receive has exceeded the threshold voltage of the memristors M1 and M5. This change in signal intensity causes the point Tac1 and the point Tac2 to output a high level of 1V. In contrast, the chemoreceptor fails to make the positive terminal voltage of the memristor M6 reach the threshold voltage due to the insufficient intensity of the received stimulation signal, and its resistance remains in a high resistance state, in a relaxed state. After subsequent logic gate circuit processing, VNoc2 is not output, and VNoc1 outputs a voltage of -1V, indicating that the baroceptor and thermoreceptor dominate in perceiving harm, and the high pressure and high temperature harm from the outside world causes the human body to produce a strong pain sensation. It should be noted that due to the difference in intensity of the external stimulation input signals VIn1 and VIn2, the corresponding pulse discharge rate presents different levels. This difference causes the pain signal VNoc received by the amygdala pathway to be significantly reduced compared to the original input signal. This phenomenon reflects the graded response characteristics of different nociceptors in the nervous system to stimulation intensity.

[0099] In Figure 8 The second set of signals is input during the second set of experiments shown. At this time, only the chemoreceptor perceives stimulation to be activated, while the baroceptor and thermoreceptor are not activated. The reason for this phenomenon is that, under this set of signals, only the signal intensity received by the memristor M6 exceeds its own threshold voltage, so the corresponding chemoreceptor generates an action potential, and after logic gate processing, the result is that VNoc1 has no output, and VNoc2 outputs a voltage of 1V, indicating that the chemoreceptor dominates in perceiving harm, and due to the action of the chemical substance, the pain is inhibited, and the human body feels a slight pain.

[0100] (2) Emotion generation:

[0101] During the experiment, the output signal of the pain grading logic circuit is input as input, transmitted and processed through the bionic emotion circuit, to observe its influence on emotion generation. Figure 9(a) And 9(b) The experimental results of emotion generation while detecting the intensity of harm are shown. As can be seen from the figure, as the intensity of the harm signal changes, the emotion circuit can produce a corresponding emotional response, and the intensity and nature of the response show a clear correlation with the input harm signal.

[0102] During the first set of experiments, the circuit detected strong injury, with the output of VNoc1 being -1V and VNoc2 having no output. VNoc1, as the input signal, activated the amygdala pathway and ultimately generated VEmo1. As shown in FIG. 9(a), VEmo1<0, indicating that the emotion generation signal is a negative emotion, which is consistent with the situation where a person shows anger and sadness when suffering from strong injury caused by high temperature and high pressure, verifying the emotion encoding mechanism of the negative valence signal. At the same time, it can be seen that the absolute value of VEmo1 weakly decreases with the input of external stimuli, indicating that the negative emotion is weakened. This is because VNoc1=-1V, which exceeds the negative threshold voltage of the memristor in the amygdala pathway, resulting in an increase in the resistance of the memristor, thereby causing the negative valence signal VEmo1 to decrease. This phenomenon simulates the process of the prefrontal cortex inhibiting the over-activation of the amygdala through glutamatergic neural projections, thereby regulating and weakening the negative emotional output.

[0103] During the second set of experiments, the circuit detected weak injury, with the output of VNoc2 being 1V and VNoc1 having no output. VNoc2, as the input signal, activated the amygdala pathway and ultimately generated VEmo2. As shown in FIG. 9(b), VEmo2>0, indicating that the emotion generation signal is a positive emotion, which is due to the inhibition of pain caused by chemical substances, such as menthol, which can activate the TRPM8 channel to produce a cooling sensation. At the same time, it can be seen that the absolute value of VEmo2 gradually increases with the input of external stimuli, indicating that the positive emotion is enhanced. This is because VNoc2=1V, which exceeds the positive threshold voltage of the memristor in the amygdala pathway, resulting in a decrease in the resistance of the memristor, thereby causing the valence signal VEmo2 to increase. This phenomenon simulates the process of pain stimulation triggering the release of endorphins by the pituitary gland, which not only inhibits the fear response of the amygdala, but also activates the dopaminergic pathway of the nucleus accumbens, thereby causing the brain to produce a pleasant feeling.

[0104] The key technical points of the present application are:

[0105] (1) Dynamic biomimetic mechanism of memristor: AIST memristor is used to replace traditional capacitors, and its nonlinear resistance state change characteristics are used to accurately simulate the dynamic membrane potential balance process of biological nociceptors, achieving threshold response and adaptive adjustment to external stimuli.

[0106] (2) Hierarchical biomimetic architecture: Based on the information processing mechanism of the biological nervous system, a three-level circuit architecture of “nociceptive encoding-pain classification-emotion mapping” is constructed, and the hierarchical signal conduction and integration functions of biological nociception are reproduced through the collaborative design of memristors and logic circuits.

[0107] (3) Multi-modal signal fusion and dynamic weight tuning: Integrate physical, temperature and chemical stimulation signals of three types of tactile nociceptor, combine with the BEL model of the amygdala pathway to realize the dynamic plasticity tuning of synaptic weights, and enhance the relevance of emotional valence signals and nociceptive input.

[0108] (4) Threshold adaptive mechanism: Through the bidirectional threshold voltage regulation of the memristor, dynamically match the intensity of external stimulation, simulate the dynamic adaptive characteristics of biological nociceptors, and avoid the limitations of traditional fixed threshold circuits.

[0109] (5) Low-power pulse signal processing: Use the non-volatile storage characteristics of the memristor and the energy efficiency advantage of the pulse neural network to reduce the dependence on traditional analog-digital conversion modules, and significantly reduce system power consumption and hardware area.

[0110] Compared with the existing technology, the beneficial effects of the present application are:

[0111] (1) Bionic and system improvement: Break through the limitations of local optimization of single device in existing technology, reproduce the hierarchical information processing mechanism of biological nervous system through system-level circuit design, and realize the deep integration of nociception and emotion generation.

[0112] (2) Dynamic adaptability enhancement: Based on the threshold dynamic adjustment function of the memristor, the dynamic adaptive characteristics of biological nociceptors can be simulated, and the response accuracy and robustness of the circuit to complex external stimuli can be significantly improved.

[0113] (3) Energy efficiency and integration optimization: Use the non-volatile storage characteristics of the memristor and the low-power advantage of the pulse neural network to eliminate the traditional ADC module, significantly reduce system power consumption, and reduce hardware area occupation.

[0114] (4) Multi-modal nociceptive grading ability: Integrate physical, temperature and chemical stimulation signals through logic circuits to realize nociceptive grading (such as severe pain, mild pain), and associate emotional valence signals (such as sadness, joy), simulate the physiological linkage mechanism of biological "pain-emotion".

[0115] (5) Biological compatibility expansion: The circuit design is compatible with memristor devices of different materials, has strong universality, and provides an expandable hardware foundation for subsequent emotional learning and memory enhancement research of brain-like intelligent systems.

[0116] The tactile nociceptor circuit based on the memristor provided in the embodiment can be applied in the following fields:

[0117] (1) Pain feedback for prosthetics has been a technical challenge. Current prosthetics only have basic touch feedback, and users often suffer secondary injuries due to lack of warning. This technology enables prosthetics to perceive dangerous stimuli such as high temperature and high pressure, just like real limbs, and alert users to avoid risks through emotional signals (such as anxiety). In particular, the simulation of chemical stimulation in this specification even provides a hardware basis for a drug delivery system - automatically releasing painkillers when pain is detected.

[0118] (2) Safe interaction with robots is another important direction. Industrial robots currently mainly rely on visual obstacle avoidance, and are not sufficient for transparent objects or sudden contact. The multi-modal perception (pressure + temperature + chemistry) and millisecond-level response of this embodiment enable robots to quickly identify dangerous contact (such as squeezing, high-temperature surfaces), and trigger different levels of emergency stop strategies through the "emotion" module. Figure 8 The micro-pain response shown is particularly valuable, which corresponds to a light warning for robots rather than direct shutdown, which can reduce false triggers on production lines.

[0119] (3) The most revolutionary possibility is in the field of neuromorphic chips. The memory-computing separation architecture of traditional AI chips leads to high energy consumption, while the mentioned memristor has a static power consumption 3-4 orders of magnitude lower than that of a capacitor, and realizes memory-computing integration. This means that bio-sensing AI that can be deployed on wearable devices for continuous operation, such as smart protective gear for athletes to monitor muscle strain risk in real time, or fireproof clothing to monitor the degree of heat damage accumulation.

[0120] Those skilled in the art will readily understand that the above description is only a preferred embodiment of the present application, and is not intended to limit the present application. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A memristor-based tactile nociceptor circuit, comprising: The application relates to a pain simulation system, which comprises the following modules: a nociceptor module for simulating the dynamic balance process of the membrane potential of a biological nociceptor by using a nonlinear resistance state change characteristic, detecting external stimuli and generating corresponding pulse signals; the external stimuli include physical, temperature and chemical stimuli; a pain grading module for integrating the pulse signals output by the nociceptor module, realizing pain grading and outputting pain signals representing slight pain and severe pain by using a logic circuit; an amygdala pathway module for dynamically generating a valence signal associated with the pain signal based on a BEL model.

2. The memristor-based tactile nociceptor circuit of claim 1, wherein, The nociceptor module comprises three different types of nociceptor circuits, i.e. tactile, temperature and chemical nociceptor circuits, each of which comprises a memristor M1, an OR gate OR1, NMOS tubes N1-N5 and PMOS tubes P1-P5. In each nociceptor circuit, the anode of the memristor M1 is connected with the drain of the PMOS tube P1 and the source of the NMOS tube N1, the cathode of the memristor M1 is connected with the gate of the PMOS tube P2, the gate of the NMOS tube N4 and the gate of the NMOS tube N5 and is connected with a fixed resistor R1; the PMOS tube P1 and the NMOS tube N1 are connected in common at the gate and are connected with the output end of the OR gate OR1; the drain of the NMOS tube N1 is connected with the source of the NMOS tube N2; the source of the PMOS tube P1 and the gate of the NMOS tube N2 are connected in common to form the input end of the nociceptor circuit; the drain of the NMOS tube N2 is connected with a-1.5V power supply; the PMOS tube P2, the NMOS tube N3 and the PMOS tube P3 form a comparator COMP1; the NMOS tube N4 and the PMOS tube P4 form a comparator COMP2; the NMOS tube N5 and the PMOS tube P5 form a comparator COMP3; the output ends of the comparator COMP1 and the comparator COMP2 are connected with the two input ends of the OR gate OR1 in correspondence; and the output end of the comparator COMP3 forms the output end of the nociceptor circuit.

3. The memristor-based tactile nociceptor circuit of claim 2, wherein, Each nociceptor circuit generates an action potential under the continuous action of corresponding external stimuli, and has four processes of depolarization, repolarization, hyperpolarization and reset in each action generation process, so as to simulate the dynamic balance process of the membrane potential of a biological nociceptor.

4. The memristor-based tactile nociceptor circuit of claim 3, wherein, Each pulse generated by each nociceptor has four stages: Na+ ions penetrate into the cell, so that the cell membrane enters a depolarization stage, the cell membrane potential starts to rise and a peak voltage is generated; When the membrane potential reaches the peak value, the Na+ channel is closed, K+ ions start to flow out of the cell, the cell membrane enters a repolarization stage and the membrane potential starts to drop; Due to the permeability of the K+ channel, the cell membrane potential drops below the resting potential, i.e. a hyperpolarization stage; In the case that there is no external stimulus input or the stimulus intensity is very low, the membrane potential returns to the resting state again.

5. The memristor-based tactile nociceptor circuit of claim 1, wherein, The pain grading module comprises a NOT gate NOT, a NAND gate NAND and AND gates AND1-AND3. The input end of the AND gate AND1 receives the pulse signal output by the nociceptor module under external physical stimulation, the other input end of the AND gate AND1 receives the pulse signal output by the nociceptor module under external temperature stimulation, and the two input ends of the AND gate AND1 are connected with the two input ends of the NAND gate NAND; the input end of the NOT gate NOT and the input end of the AND gate AND3 both receive the pulse signal output by the nociceptor module under external chemical stimulation, the output end of the NOT gate NOT is connected with the input end of the AND gate AND2, the other input end of the AND gate AND2 is connected with the output end of the AND gate AND1, the other input end of the AND gate AND3 is connected with the output end of the NAND gate NAND, and the output end of the AND gate AND2 and the output end of the AND gate AND3 constitute the two output ends of the pain grading module.

6. The memristor-based tactile nociceptor circuit of claim 1, wherein, The amygdala pathway module comprises two amygdala pathway circuits, each of which comprises a memristor M2-M4 and an operational amplifier OP1 and OP2. In each amygdala access circuit, the memristor M2 and the memristor M3 are connected in series, the positive electrode of the memristor M2 is the input end of the amygdala access circuit, the negative electrode of the memristor M2 is connected with the fixed resistor R9 and the capacitor C1; the negative electrode of the memristor M3 is connected with the inverting input end of the operational amplifier OP1, the non-inverting input end of the operational amplifier OP1 is grounded, and the output end of the operational amplifier OP1 is connected with the fixed resistor R 10 connected with the inverting input end of the operational amplifier OP1, and the output end of the operational amplifier OP1 is connected with the fixed resistor R 11 connected with the inverting input end of the operational amplifier OP2, the non-inverting input end of the operational amplifier OP2 is grounded, and the output end of the operational amplifier OP2 is connected with the fixed resistor R 12 connected with the inverting input end of the operational amplifier OP2; the positive electrode of the memristor M4 is connected with the output end of the operational amplifier OP2, and the negative electrode of the memristor M4 is connected with the fixed resistor R 13 , and is the output end of the amygdala access circuit.

7. The memristor-based tactile nociceptor circuit of claim 6, wherein, The value of the potency signal depends on the pain signal and the memristive value of the memristors M2 and M3.

8. Use of a memristor-based tactile nociceptor circuit according to any one of claims 1 to 7, characterized in that, It is applied to the fields of prosthetics, safe interaction of robots and neuromorphic chips.

9. The application of a memristor-based tactile nociceptor circuit according to claim 8, wherein, The neuromorphic chip is deployed on a wearable device.

10. The application of a memristor-based tactile nociceptor circuit as defined in claim 9, wherein, The wearable device comprises an athlete's intelligent protective gear and a firefighter's uniform. When deployed on the athlete's intelligent protective gear, it is used for real-time monitoring of the athlete's muscle strain risk. When deployed on the firefighter's uniform, it is used for monitoring the degree of cumulative thermal injury.

Images