A method and system for biomimetic multi-sensory-motor behavior of caenorhabditis elegans

Abstract

The application discloses a Caenorhabditis elegans bionic multi-perception-motion behavior method and system, and belongs to the field of artificial pulse neural network control. A pulse neural network model for changing the motion behavior of Caenorhabditis elegans by sensing changes in salt concentration, temperature and tactile stimulation position in the environment is built, the pulse neural network model comprises a bionic forward loop, a backward loop, a salt-seeking loop, a temperature-seeking loop and a tactile loop, and the bionic multi-perception-motion behavior of Caenorhabditis elegans is simulated. The salt-seeking behavior, the temperature-seeking behavior and the touch-avoiding behavior of Caenorhabditis elegans can be better matched, and a complex multi-perception-motion behavior mode can be realized by using a small number of neurons.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial spiking neural network control, in particular to a Caenorhabditis elegans bionic multi-sensing-motion behavior method and system. BACKGROUND

[0002] Caenorhabditis elegans is a small worm living in soil, which has been widely used as a powerful model organism since the early 1970s. The study of the nervous system of the nematode has a long history, and it is the only organism whose neural connections have been completely constructed. They have only 302 neurons, but have rich behavioral characteristics. Including sinusoidal motion, temperature perception, chemical perception, tactile perception, learning adaptation and other behaviors. It has an important enlightening effect on the study and revelation of the neural and brain neural connection and operation mechanism of higher organisms, and also has an enlightening effect on multi-modal multi-task sensing applications, such as sweeping robot, autonomous driving and other application scenarios.

[0003] In recent years, with the development of the field of artificial intelligence, constructing the neural connection group of model organisms and simulating its basic behavior, outputting the weights of the connections, provides a new technical means for in-depth understanding of the complex behavior ability of simple organisms, and also provides a more flexible verification means for neuroscience in addition to biological experiments. However, traditional neural networks are based on mathematical models and have no obvious physical meaning. Spiking neural networks have the characteristics of biological neural simulation and can more realistically simulate the state and behavior of nematode neurons.

[0004] Therefore, using spiking neural networks to study and imitate the motion and behavior principles of model organisms can be applied to bionic robots, path planning, obstacle avoidance navigation and other motion control. SUMMARY

[0005] The purpose of the present application is to provide a Caenorhabditis elegans bionic multi-sensing-motion behavior method and system, which can simulate the Caenorhabditis elegans bionic multi-sensing-motion behavior, better match the salt-seeking, temperature-seeking and touch-avoiding behavior of Caenorhabditis elegans, and realize complex multi-sensing-motion behavior patterns with fewer neurons.

[0006] To achieve the above purpose, the present application provides the following scheme:

[0007] A Caenorhabditis elegans bionic multi-sensing-motion behavior method, comprising:

[0008] constructing an environment model in which the Caenorhabditis elegans is located; the environmental parameters in the environment model include salt concentration, temperature and tactile stimulation position;

[0009] constructing a morphological structure model of the Caenorhabditis elegans;

[0010] According to the neural structure of Caenorhabditis elegans, a pulse neural network model for changing the motion behavior of Caenorhabditis elegans by sensing the change of salt concentration, temperature and tactile stimulation position in the environment is built; the pulse neural network model comprises a bionic forward loop, a backward loop, a salt-seeking loop, a temperature-seeking loop and a tactile loop;

[0011] The morphological structure model of Caenorhabditis elegans is placed in the environment model for simulation;

[0012] During the simulation process, if Caenorhabditis elegans senses the change of the environment, the current position of Caenorhabditis elegans senses the changed environmental parameters and inputs the pulse neural network model to control the morphological structure model to stretch or shrink, so that the body of Caenorhabditis elegans produces a bending shape, and the motion control of Caenorhabditis elegans in the simulation environment is realized.

[0013] A Caenorhabditis elegans bionic multi-perception-motion behavior system comprises an environment module, a nematode neural network calculation module, a nematode physical information calculation module and a nematode simulation visualization module;

[0014] The environment module is used for simulating the environment in which Caenorhabditis elegans is located; the environmental parameters include salt concentration, temperature and tactile stimulation position;

[0015] The nematode neural network calculation module is used for taking the changed environmental parameters sensed by the current position of Caenorhabditis elegans as input quantities during the simulation process of the morphological structure model of Caenorhabditis elegans in the environment model, and generating a stretching and shrinking signal to the nematode physical information calculation module;

[0016] The nematode physical information calculation module is used for controlling Caenorhabditis elegans to stretch or shrink according to the stretching and shrinking signal, and feeding back the stretching and shrinking information of the abdomen and back of Caenorhabditis elegans to the nematode neural network calculation module;

[0017] The nematode neural network calculation module is also used for continuously generating a stretching and shrinking signal according to the feedback stretching and shrinking information and outputting to the nematode physical information calculation module to realize the interaction with the nematode physical information calculation module;

[0018] The nematode simulation visualization module is used for visualizing the motion form of Caenorhabditis elegans, the interaction between Caenorhabditis elegans and the environment and the neural state of Caenorhabditis elegans in real time.

[0019] According to the specific embodiments of the present application, the following technical effects are provided:

[0020] The application discloses a Caenorhabditis elegans bionic multi-perception-motion behavior method and system. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments will be briefly introduced below. Obviously, the drawings in the following description only constitute some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor under the premise of the drawings.

[0022] Figure 1 A flowchart of a Caenorhabditis elegans bionic multi-perception-motion behavior method provided by the embodiment of the present application is provided.

[0023] Figure 2 A morphological structure model diagram of Caenorhabditis elegans provided by the embodiment of the present application is provided.

[0024] Figure 3 A structural schematic diagram of a fragment provided by the embodiment of the present application is provided.

[0025] Figure 4 A loop oscillation process schematic diagram of a central pattern generator provided by the embodiment of the present application is provided.

[0026] Figure 5 A salt-seeking loop schematic diagram provided by the embodiment of the present application is provided.

[0027] Figure 6 A temperature-seeking loop schematic diagram provided by the embodiment of the present application is provided.

[0028] Figure 7 A tactile loop schematic diagram provided by the embodiment of the present application is provided.

[0029] Figure 8 A temperature-seeking behavior simulation path schematic diagram provided by the embodiment of the present application is provided.

[0030] Figure 9 A genetic algorithm training model schematic diagram provided by the embodiment of the present application is provided.

[0031] Figure 10 A genetic algorithm training result schematic diagram provided by the embodiment of the present application is provided.

[0032] Figure 11 A structural schematic diagram of a Caenorhabditis elegans bionic multi-perception-motion behavior system provided by an embodiment of the present application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort fall within the protection scope of the present application.

[0034] The present application aims to provide a Caenorhabditis elegans bionic multi-perception-motion behavior method and system, which can simulate the Caenorhabditis elegans bionic multi-perception-motion behavior, better match the salt-seeking behavior, temperature-seeking behavior and touch-avoiding behavior of the Caenorhabditis elegans, and realize a complex multi-perception-motion behavior pattern with a small number of neurons.

[0035] In order to make the above-mentioned objects, features and advantages of the present application more apparent and comprehensible, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0036] Since the perception-motion behavior of the nematode is controlled only by the nervous system, the present application adopts a method of pulse neural network control to realize the perception-motion behavior. Based on the strong learning ability of the pulse neural network, as long as the adopted neural network can effectively approximate the constructed nonlinear logic function, the neural network can complete the perception-motion behavior, and then be extended to a robot navigation system controlled by the neural network.

[0037] Embodiment one

[0038] As shown in Figure 1 The present application provides a Caenorhabditis elegans bionic multi-perception-motion behavior method, which comprises the following steps:

[0039] Step 1: Construct an environment model in which the Caenorhabditis elegans is located; the environment parameters in the environment model include salt concentration, temperature and touch stimulation position.

[0040] A coordinate system for simulating the salt concentration environment, temperature gradient and touch stimulation position of the Caenorhabditis elegans is established, and the salt concentration distribution, temperature gradient and touch stimulation position are set in the established coordinate system. The salt concentration, temperature and touch stimulation position can be self-defined, and the initial value is a random point in the plane. The salt concentration distribution in the coordinate system is modeled by linear distribution, the temperature condition is modeled by Gaussian distribution, and the touch stimulation position is set as a coordinate point.

[0041] The function of the linear distribution is: Φ(t)=c*(ΔC-ΔC0)+d.

[0042] In the formula, Φ(t) represents the deflection angle, ΔC represents the salt concentration difference between time t and t-1, ΔC0 represents the concentration difference between the salt concentration at time t and the optimum salt concentration C0, and c and d are the first and second constants, respectively.

[0043] Step 2: Construct a morphological model of Caenorhabditis elegans.

[0044] Based on the muscular anatomy of *C. elegans*, a morphological model of *C. elegans* was constructed, consisting of multiple segments, axially rigid rods, spring cords, and stretch receptors. Figure 2 As shown.

[0045] The morphological model comprises 24 muscle blocks and 48 segments. The 24 muscle blocks are divided into 6 head muscle blocks and 18 body muscle blocks. Each muscle block connects to two segments; adjacent muscle blocks have overlapping segments. Each segment includes a stretch receptor, two axial stiffness rods, two transverse spring cables, and two diagonal spring cables. The two axial stiffness rods are positioned along the longitudinal extension of the *C. elegans* body. The two axial stiffness rods and two transverse spring cables form a quadrilateral, with the two diagonal spring cables positioned along the two diagonals of the quadrilateral. The stretch receptors are used to acquire stretching information from the ventral and dorsal sides of the *C. elegans*, converting it into electrical output. Muscle dynamics are primarily output to the transverse spring cables, thereby controlling the stretching and contraction of each segment, causing the nematode to exhibit an overall sinusoidal bending morphology.

[0046] The structure of each segment is as follows: Figure 3 As shown. Figure 3 In the diagram, L represents the transverse rigid bar, D represents the diagonal spring cable, CoM represents the axial rigid bar, and p represents the focus of the three lines. The superscript D indicates dorsal, the superscript V indicates Ventra, and the subscript i indicates the i-th segment. (x, y, φ) represent the coordinates and angles.

[0047] The specific correspondence between muscle blocks and segments is as follows:

[0048] Muscle_to_Segment: [[0, 1], 2], [[0, 1], 3], [[1, 2], 4], [[1, 2], 5], [[2, 3], 6], [[2, 3], 7], [[3, 4], 8], [[3, 4], 9], [[4, 5], 10], [[4, 5], 11], [[5, 6], 12], [[5, 6], 13], [[6, 7], 14], [[6, 7], 15], [[7, 8], 16], [[7, 8], 17], [[8, 9], 18], [[8, 9], 19], [[9, 10], 20], [[9, 10], 21], [[10, 11], 22], [[10, 11], 23], [[11, 12], 24], [[11, 12], 25], [[12, 13], 26], [[12, 13], 27], [[13, 14], 28], [[13, 14], 29], [[14, 15], 30], [[14, 15], 31], [[15, 16], 32], [[15, 16], 33], [[16, 17], 34], [[16, 17], 35], [[17, 18], 36], [[17, 18], 37], [[18, 19], 38], [[18, 19], 39], [[19, 20], 40], [[19, 20], 41], [[20, 21], 42], [[20, 21], 43], [[21, 22], 44], [[21, 22], 45], [[22, 23], 46], [[22, 23], 47]. For example, [0, 1], 2] means that the 0th and 1st segments are connected to the 2nd muscle block.

[0049] The specific corresponding connection relationship between the segment and the motor neuron module is as follows:

[0050] Segment_to_Motor_Neuron_Module: [[6, 7, 8, 9, 10, 11], 0], [[10, 11, 12, 13, 14, 15], 1], [[14, 15, 16, 17, 18, 19], 2], [[18, 19, 20, 21, 22, 23], 3], [[22, 23, 24, 25, 26, 27], 4], [[26, 27, 28, 29, 30, 31], 5]. For example, [[6, 7, 8, 9, 10, 11], 0] means that the 6th, 7th, 8th, 9th, 10th, and 11th segments are connected to the 0th motor neuron module.

[0051] Each muscle block includes ventral and dorsal input and activation, and the muscle formula is

[0052]

[0053] wherein, represents the activation of a muscle block; represents the total current driving the muscle block; τ M represents the length of time, τ M = 100 ms; t represents time; the subscript M represents the total number of muscle blocks, M = 24; the subscript m represents the segment number, m = [1, 48]; the superscript k represents the ventral or dorsal side, k belongs to {dorsal, ventral}, dorsal represents the dorsal side, and ventral represents the ventral side.

[0054] Step 3: According to the neural structure of C. elegans, a spiking neural network model for changing the movement behavior of C. elegans in response to changes in salt concentration, temperature, and tactile stimulus position in the environment is built; the spiking neural network model includes a bionic forward loop, a backward loop, a salt-seeking loop, a temperature-seeking loop, and a tactile loop.

[0055] The spiking neural network model includes: a perception neuron, an intermediate neuron, and a motor neuron.

[0056] (1) Forward loop

[0057] Referring to Figure 4 , the forward loop only includes motor neurons; the motor neurons of the forward loop all include SMDV neurons, RMDV neurons, SMDD neurons, and RMDD neurons.

[0058] The forward loop uses neurons that are bistable neurons, and the formula is:

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065] wherein, C mem is the membrane capacitance, is the continuous value membrane potential of the nth motor neuron on the ventral or dorsal side;

[0066] is the leakage current, G mem and V rest are the membrane leakage conductance and the reversal potential, respectively;

[0067] G is a state-dependent self-exciting current, act Gmax is the maximum conductance, k act and V act are first and second activation function parameters, respectively;

[0068] Gstretch is the stretch receptor current, summed over m body segments; A n is a segment-specific compensation parameter, G SR,n is a conductance parameter that increases linearly from head to tail to compensate for the reduction in the amplitude of the worm’s undulations due to the gradient of muscle efficacy; N out is the number of remaining posterior segments; s is the number of segments that perceive stretch, s = min{M; N SR + (n - 1)N out}, s is minimized over M = 48 body segments and N SR + the number of posterior output segments. Gstretch is the stretch receptor current, summed over m body segments; A L 0L,m is the segment residual length, is the current length of the mth segment’s kth edge; R is the radius of the body curvature; Modify the dorsal stretch receptors to ensure that the worm can still move straight in the presence of neural circuit asymmetry.

[0069] G is the persistent input current from the AVB neuron, G AVB Gsyn is the conductance of the synaptic gap, V AVB V is the membrane potential of the AVB neuron, Ibias is the bias current;

[0070] G is the inhibitory input current, G GABA Gmax is the maximum synaptic conductance for neural inhibition, k GABA V is the first activation parameter, V is the membrane potential of the nth neuron in the dorsal end, v 0,GABA V is the second activation parameter.

[0071] Three conditions for the forward loop:

[0072] 1. Forward signal - Central Pattern Generator (CPG)

[0073] 2. Alternating activation of ventral and dorsal muscle cells (mutually inhibitory loop structure)

[0074] 3. Head muscle cell activation precedes tail (CPG signal from head to tail)

[0075] CPG (Central Pattern Generator) loop oscillation process:

[0076] 1. Initial: SMDV neuron and RMDD neuron activation, resulting in muscle activation, nematode body (stretch receptor) sends compression signal.

[0077] 2. SMDD neuron receives stretch receptor signal, gradually activates, while inhibiting SMDV neuron, activating RMDV neuron.

[0078] 3. After RMDV neuron activation, inhibit RMDD neuron, resulting in muscle cell deactivation, nematode body (stretch receptor) sends stretch signal.

[0079] 4. SMDD neuron receives stretch signal, gradually deactivates, inhibition weakens, resulting in SMDV neuron and RMDD neuron reactivation.

[0080] Ventral nerve cord loop (VNC) loop process:

[0081] 1. Assuming VB neuron activation, VB neuron will activate DD neuron, VB neuron stimulates ventral muscle cells while inhibiting dorsal cells.

[0082] 2. Similarly, DB neuron activation will stimulate dorsal cells while inhibiting ventral muscle cells.

[0083] (2) Retreating loop

[0084] The retreating loop is the reverse process of the advancing loop.

[0085] (3) Salt-seeking loop

[0086] Reference Figure 5 The perception neurons of the salt-seeking loop include ASEL neurons and ASER neurons, intermediate neurons include AIYL neurons, AIYR neurons, AIZL neurons and AIZR neurons, and motor neurons include SMBVL neurons, SMBVR neurons, SMBDL neurons and SMBDR neurons.

[0087] The salt-seeking loop makes the nematode move in the direction of ascending the salt gradient in the environment, and when the NaCL concentration is too high, the salt taxis is reversed, and the nematode moves in the direction of descending the gradient.

[0088] Salt-seeking loop related calculation formula:

[0089] Nacl perception neuron:

[0090]

[0091]

[0092]

[0093] where V ON (t) is the membrane potential of the ON neuron at time t, d(t) is the activation value, V OFF (t) is the membrane potential of the ON neuron at time t, c(t) is the concentration at time t, N and M' are two interval durations of the average concentration.

[0094] Interneurons and motor neurons:

[0095]

[0096]

[0097]

[0098] V PG = sin(2πt / T)

[0099] where τ is the time constant; V is the membrane potential; w is the chemical synaptic strength parameter between neurons; g is the conductance parameter; σ is the synaptic potential or output of the neuron; θ is the bias term of the range of sensitivity of the displacement output function of the neuron; T is the time period; the subscripts AIYL, AIYR, ON, OFF, AIZL, AIZR, SMBDL, PG, AIY, and AIZ represent the AIYL neuron, the AIYR neuron, the ON neuron, the OFF neuron, the AIZL neuron, the AIZR neuron, the SMBDL neuron, the PG neuron, the AIY neuron, and the AIZ neuron, respectively.

[0100] (4) Thermotactic circuit

[0101] Referring to Figure 6 , the sensory neurons of the thermotactic circuit include the AFD neuron and the AWC neuron, the interneurons include the AIY neuron and the AIZ neuron, and the motor neurons include the RIA neuron.

[0102] This function is that when the sensory neurons AFD and AWC perceive the temperature, the nematode is driven to approach or avoid, so as to find a suitable temperature environment.

[0103] Figure 8 It is a simulation path of the thermotactic behavior, which shows that the nematode can move along a suitable temperature interval. Figure 8 A 2D map is shown, and both the horizontal coordinate and the vertical coordinate are the size of the map.

[0104] (5) Tactile circuit

[0105] Referring to Figure 7 , the advancing sensory neurons in the tactile loop include ALM neurons and AVM neurons, the advancing interneurons include AVB neurons and PVC neurons, and the advancing motor neurons include DB neurons and VB neurons. The retreating sensory neurons in the tactile loop include PVD neurons and PLM neurons, the retreating interneurons include AVA neurons and AVD neurons, and the retreating motor neurons include DA neurons and VA neurons.

[0106] When the head of the nematode encounters a touch, the tactile sensory neurons are activated, which further drives the nematode to retreat. When the tail of the nematode encounters a touch, the tactile sensory neurons are activated, which further drives the nematode to advance.

[0107] After the spiking neural network model is constructed, the loops in the model are trained: by introducing Figure 9 a genetic algorithm shown in FIG. 6, the nematodes capable of normal advancing or salt-seeking are screened (evolved) from the population.

[0108] Figure 9 The genetic algorithm training model provided for the embodiments of the present application screens (evolves) the nematodes capable of normal advancing or salt-seeking from the population. Figure 10 The genetic algorithm training result provided for the embodiments of the present application trains the advancing neuron output. Figure 10 The abscissa represents the round, and the ordinate represents the firing value of the four neurons.

[0109] Step 4: placing the morphological structure model of the C. elegans in the environment model for simulation.

[0110] Step 5: if the C. elegans senses a change in the environment during the simulation, inputting the changed environmental parameters sensed by the current position of the C. elegans into the spiking neural network model, controlling the morphological structure model to stretch or contract, and making the body of the C. elegans produce a curved shape, so as to realize the motion control of the C. elegans in the simulation environment.

[0111] For example, the specific process of controlling the motion of the C. elegans is as follows:

[0112] ① presetting the serial number i as 1;

[0113] ② inputting the changed environmental parameters sensed by the current position of the C. elegans into the spiking neural network model, outputting the stretch signal to the first muscle block in the morphological structure model; the changed environmental parameters are one or more of the salt concentration, the temperature, and the tactile stimulation position;

[0114] ③ the first muscle block transmits the stretch signal to the corresponding segment, the corresponding segment stretches or contracts according to the stretch signal, and the stretch receptors in the corresponding segment obtain the stretch information of the abdomen and back of the C. elegans and convert it into an electric current feedback to the motor neurons of the pulse neural network model;

[0115] ④ the motor neurons control the stretch activity of the first + i muscle block and the morphology of the corresponding segment, and the stretch receptors in the corresponding segment of the first + i muscle block obtain the stretch information of the abdomen and back of the C. elegans and convert it into an electric current feedback to the motor neurons of the advance loop or the retreat loop in the pulse neural network model;

[0116] ⑤ increase the value of i by 1 and return to step ④, until the value of i is equal to the total number of muscle blocks in the morphological structure model, stop, complete the control of the C. elegans body to produce a bending shape.

[0117] The following illustrates the actual simulation operation and control process of the C. elegans:

[0118] After placing the morphological structure model of the C. elegans in the environment model, the central pattern generator starts to control the morphological structure model to move forward.

[0119] The salt-seeking loop makes the nematode move in the direction of ascending the salinity gradient in the environment, and when the concentration of NaCL is too high, the salt-seeking is reversed, and the nematode moves in the direction of descending the gradient.

[0120] When the perception neurons AFD and AWC of the thermotactic loop sense the temperature, they will drive the nematode to approach or avoid, so as to find a suitable temperature environment.

[0121] During the simulation process, the user clicks on any position of the morphological structure model to simulate that the C. elegans has been stimulated by the tactile stimulation of the obstacle.

[0122] When the head of the nematode encounters a touch, the perception neurons of the tactile loop are activated, and the motor neurons of the tactile loop output a contraction signal to the first muscle block located at the tail of the morphological structure model. The first muscle block transmits the contraction signal to the corresponding segment, the corresponding segment contracts according to the contraction signal, and the stretch receptors in the corresponding segment obtain the contraction information of the abdomen and back of the C. elegans and convert it into an electric current feedback to the motor neurons of the central pattern generator. The motor neurons of the central pattern generator control the contraction activity of the second muscle block and the morphology of the corresponding segment, and the motor neurons of the central pattern generator successively control the contraction of the third, fourth, … muscle blocks, from the tail to the head of the nematode, to complete the retreat of the C. elegans to avoid the obstacle.

[0123] When the tail of the nematode meets the touch, the sensory neurons of the tactile loop are activated, and the motor neurons of the tactile loop output the stretch signal to the first muscle block located at the head of the morphological structure model, and the first muscle block transmits the stretch signal to the corresponding segment, and the corresponding segment stretches according to the stretch signal, and the stretch receptors in the corresponding segment obtain the stretch information of the dorsal and ventral of the nematode, and are converted into an electric current and fed back to the motor neurons of the central pattern generator. The motor neurons of the central pattern generator control the stretch activity of the second muscle block and the shape of the corresponding segment, and in turn, the motor neurons of the central pattern generator control the stretch of the third, fourth, … muscle blocks, from the head to the tail of the nematode, to complete the forward obstacle avoidance of the nematode.

[0124] The superimposed environment module input makes the sensory neurons on the ventral and dorsal sides receive different signals, and when the dorsal side contracts, the ventral side stretches, or when the dorsal side stretches, the ventral side contracts, thereby producing different body segment stretching and contraction to make the nematode deflect in different directions.

[0125] The present application breaks through the theoretical limitations of traditional robots, and uses the powerful learning and prediction ability of artificial neural networks to process and predict the motion at the next moment based on the external sensory information collected by the sensor, thereby realizing the avoidance and turning motion ability of the crawler robot. It provides a good reference for improving the motion control mode, autonomous control ability and complex environment adaptation ability of future robots.

[0126] The robot established by the present application can be applied in the following four fields or other applicable fields: 1) automatic driving road detection; 2) rescue of victims in complex and dangerous environments; 3) examination of human organs such as stomach and intestines in medicine; 4) inspection and repair of industrial pipelines.

[0127] Embodiment two

[0128] Embodiment two provides a Caenorhabditis elegans bionic multi-sensing-motion behavior system, as shown in the figure, which comprises an environment module, a nematode neural network calculation module, a nematode physical information calculation module and a nematode simulation visualization module. Figure 11

[0129] The environment module is used to simulate the environment in which the Caenorhabditis elegans is located; the environment parameters include salt concentration, temperature and tactile stimulation position.

[0130] The nematode neural network calculation module is used to generate stretch signals to the nematode physical information calculation module during the simulation process of the morphological structure model of the Caenorhabditis elegans in the environment model, taking the changed environment parameters in the current position of the Caenorhabditis elegans as input quantities.

[0131] ​The nematode physical information calculation module is configured to control the Caenorhabditis elegans to stretch or contract according to the stretch signal, and feed back the stretch information of the Caenorhabditis elegans to the nematode neural network calculation module.

[0132] The nematode neural network calculation module is further configured to continue to generate the stretch signal according to the feedback stretch information, and output the stretch signal to the nematode physical information calculation module to realize the interaction with the nematode physical information calculation module.

[0133] The nematode simulation visualization module is configured to visualize the motion form of the Caenorhabditis elegans, the interaction between the Caenorhabditis elegans and the environment, and the neural state of the Caenorhabditis elegans in real time.

[0134] The system mainly comprises: 1) an environment perception module, which is configured to set the isotherm distribution of the environment temperature, the salinity distribution and the position of the tactile stimulation; 2) a nematode physical information calculation module, which is configured to set the form, mechanical structure and muscle power of the nematode. The output signal of the neural circuit is obtained and transmitted to the nematode muscle receptor to drive the change of the mechanical structure and form of the nematode; 3) a nematode neural network calculation module, which is configured to construct the forward circuit, the backward circuit, the salt-seeking circuit and the tactile circuit of the simulation nematode. The motion behavior can be changed in response to the changes of the environment salinity and the tactile and obstacles; 4) a nematode simulation visualization module: double threads including the simulation construction of the nematode motion and the environment and the visualization of the nematode neural state; the system has important theoretical significance for the research on the essence of biological perception-motion behavior. At the same time, the system has important guiding significance for the research on the neural network construction, path planning, motion strategy selection and deflection motion control of robots.

[0135] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiments, the description is relatively simple because it corresponds to the method disclosed in the embodiments. The relevant parts can be referred to the description of the method.

[0136] The principles and implementation manners of the present application are described by using specific examples. The above description of the embodiments is only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, the specific implementation manners and application scope can be changed according to the idea of the present application. In conclusion, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method of biomimetic multi-sensory-motor behavior of Caenorhabditis elegans, characterized in that, The application relates to a simulation method for controlling movement of a Caenorhabditis elegans in a simulation environment. The method comprises the following steps: constructing an environment model in which the Caenorhabditis elegans exists; environment parameters in the environment model include salt concentration, temperature and position of a touch stimulus; constructing a morphological structure model of the Caenorhabditis elegans; according to a neural structure of the Caenorhabditis elegans, constructing a spiking neural network model for changing movement behavior of the Caenorhabditis elegans in response to changes in salt concentration, temperature and position of a touch stimulus in the environment; the spiking neural network model comprises a bionic forward loop, a backward loop, a salt-seeking loop, a temperature-seeking loop and a touch loop; placing the morphological structure model of the Caenorhabditis elegans in the environment model for simulation; 2. The C. elegans biomimetic multi-sensory-motor behavior method of claim 1, wherein, if the Caenorhabditis elegans senses changes in the environment during the simulation, inputting environment parameters sensed by the Caenorhabditis elegans at a current position into the spiking neural network model, controlling the morphological structure model to stretch or contract, and making the body of the Caenorhabditis elegans have a bending shape, so as to realize movement control of the Caenorhabditis elegans in the simulation environment. The salt concentration in the environment model is linearly distributed, and the temperature is distributed in a Gaussian distribution; a function of the linear distribution is Phi (t) = c * (Delta C - Delta C0) + d; 3. The C. elegans biomimetic multi-sensory-motor behavior method of claim 1, wherein, in the formula, Phi (t) represents a deflection angle value, Delta C represents a salt concentration difference between t and t-1, Delta C0 represents a concentration difference between a salt concentration at t and an optimal salt concentration C0, and c and d are respectively first and second constants. the morphological structure model comprises 24 muscle blocks and 48 segments; the 24 muscle blocks are divided into 6 head muscle blocks and 18 body muscle blocks; one muscle block is connected with two segments; segments connected by adjacent muscle blocks have overlaps; each segment comprises a stretch receptor, two axial rigid rods, two transverse spring cables and two diagonal spring cables; the two axial rigid rods are arranged along a longitudinal extension direction of the body of the Caenorhabditis elegans; the two axial rigid rods and the two transverse spring cables form a quadrilateral, and the two diagonal spring cables are arranged on two diagonal lines of the quadrilateral; 4. The C. elegans biomimetic multi-sensory-motor behavior method of claim 3, wherein, the stretch receptor is used for acquiring stretch information of the abdomen and back of the Caenorhabditis elegans and converting the stretch information into an electric current output. the method for inputting the environment parameters sensed by the Caenorhabditis elegans at the current position into the spiking neural network model, controlling the morphological structure model to stretch or contract, and making the body of the Caenorhabditis elegans have a bending shape specifically comprises the following steps: presetting a serial number i as 1; inputting the environment parameters sensed by the Caenorhabditis elegans at the current position into the spiking neural network model, outputting a stretching and contracting signal to a first muscle block in the morphological structure model, and the environment parameters being one or more of the salt concentration, the temperature and the position of the touch stimulus; the first muscle block transmits the stretching and contracting signal to a corresponding segment, the corresponding segment stretches or contracts according to the stretching and contracting signal, the stretch receptor in the corresponding segment acquires stretch information of the abdomen and back of the Caenorhabditis elegans, and the stretch information is converted into an electric current and fed back to a motor neuron of the spiking neural network model. The motor neuron controls the stretching and contracting of the 1+i muscle block and the shape of the corresponding segment, and the stretch receptor in the corresponding segment of the 1+i muscle block obtains the stretching information of the C. elegans body and converts it into an electric current feedback to the motor neuron of the forward loop or the backward loop of the spiking neural network model; Increase the value of i by 1 and return to the step "the motor neuron controls the stretching and contracting of the 1+i muscle block and the shape of the corresponding segment, and the stretch receptor in the corresponding segment of the 1+i muscle block obtains the stretching information of the C. elegans body and converts it into an electric current feedback to the motor neuron of the forward loop or the backward loop of the spiking neural network model", until the value of i is equal to the total number of muscle blocks in the shape structure model, and the control of the C. elegans body producing a bending shape is completed.

5. The C. elegans biomimetic multi-sensory-motor behavior method of claim 3, wherein, The activation formula of the muscle block is wherein represents activation of a muscle segment; represents total current driving the muscle segment; τ M represents a length of time; t represents time; subscript M represents total number of muscle segments, M = 24; subscript m represents segment number, m = [1, 48]; superscript k represents ventral or dorsal.

6. The C. elegans biomimetic multi-sensory-motor behavior method of claim 5, wherein, The spiking neural network model comprises perception neurons, intermediate neurons and motor neurons; The forward loop and the backward loop only comprise motor neurons; the motor neurons of the forward loop and the backward loop comprise SMDV neurons, RMDV neurons, SMDD neurons and RMDD neurons; The perception neurons of the salt-seeking loop comprise ASEL neurons and ASER neurons, the intermediate neurons comprise AIYL neurons, AIYR neurons, AIZL neurons and AIZR neurons, and the motor neurons comprise SMBVL neurons, SMBVR neurons, SMBDL neurons and SMBDR neurons; The perception neurons of the temperature-seeking loop comprise AFD neurons and AWC neurons, the intermediate neurons comprise AIY neurons and AIZ neurons, and the motor neurons comprise RIA neurons; The forward perception neurons of the touch loop comprise ALM neurons and AVM neurons, the forward intermediate neurons comprise AVB neurons and PVC neurons, and the forward motor neurons comprise DB neurons and VB neurons; The backward perception neurons of the touch loop comprise PVD neurons and PLM neurons, the backward intermediate neurons comprise AVA neurons and AVD neurons, and the backward motor neurons comprise DA neurons and VA neurons.

7. The C. elegans biomimetic multi-sensory-motor behavior method of claim 6, wherein, The forward loop uses bistable neurons, and the calculation formula of the forward loop is wherein C mem is a membrane capacitance, is a continuous value membrane potential of the ventral or dorsal nth motor neuron; For the leakage current, G mem and V rest are the membrane leak conductance and reversal potential, respectively; G is a state-dependent self-excited current act k is the maximum conductance act and V act are first and second activation function parameters, respectively; A for stretching the susceptor current n G for a compensation parameter for the segment SR,n N for a conductance parameter out s for the number of segments perceived to be stretched, A for an effective stretching susceptor activation function; is the input current from the AVB neuron, G AVB is the conductance of the synaptic gap, V AVB is the membrane potential of the AVB neuron, is the bias current; G is the input current to be suppressed, GABA k is the maximum synaptic conductance for the neural suppression, GABA a is the first activation parameter, v is the membrane potential of the nth neuron at the back end, 0,GABA b is the second activation parameter.

8. The C. elegans biomimetic multi-sensory-motor behavior method of claim 6, wherein, The three conditions of the forward loop comprise that the central pattern generator sends a forward signal, the ventral and dorsal muscle cells are activated alternately, and the head muscle cells are activated earlier than the tail muscle cells; The loop oscillation process of the central pattern generator is: The SMDV neurons and the RMDD neurons are activated, causing muscle activation and the C. elegans body to send a compression signal; The SMDD neurons receive the compression signal, gradually activate, and inhibit the SMDV neurons, and activate the RMDV neurons; After the RMDV neurons are activated, the RMDD neurons are inhibited, causing the muscle cells to be inactivated, and the C. elegans body sends a stretching signal; The SMDD neurons receive the stretching signal, gradually inactivate, and the inhibition is weakened, causing the SMDV neurons and the RMDD neurons to be reactivated; The alternating activation loop process of the ventral and dorsal muscle cells is as follows: If the VB neuron is activated, the VB neuron activates the DD neuron, and the VB neuron stimulates the ventral muscle cell while inhibiting the dorsal cell; If the DB neuron is activated, the DB neuron stimulates the dorsal cell while inhibiting the ventral muscle cell.

9. The C. elegans biomimetic multi-sensory-motor behavior method of claim 6, wherein, The calculation formula of the sensory neuron in the salt-seeking loop is where VON(t) is the membrane potential of the ON neuron at time t, d(t) is the activation value, V OFF where VON(t) is the membrane potential of the ON neuron at time t, c(t) is the concentration at time t, N and M' are two intervals of the average concentration duration; The calculation formula of the intermediate neuron and the motor neuron in the salt-seeking loop is V PG = sin(2πt / T); In the formula, τ is a time constant, V is a membrane potential, w is a chemical synaptic strength parameter between neurons, g is a conductance parameter, σ is a synaptic potential or output of a neuron, θ is a bias term of a displacement output function sensitivity range of a neuron, T is a time period, and the subscripts AIYL, AIYR, ON, OFF, AIZL, AIZR, SMBDL, PG, AIY, and AIZ represent AIYL neurons, AIYR neurons, ON neurons, OFF neurons, AIZL neurons, AIZR neurons, SMBDL neurons, PG neurons, AIY neurons, and AIZ neurons, respectively.

10. A Caenorhabditis elegans biomimetic multi-sensory-motor behavior system, characterized in that, The simulation system comprises: an environment module, a nematode neural network calculation module, a nematode physical information calculation module, and a nematode simulation visualization module; the environment module is used to simulate the environment in which the C. elegans is located; the environment parameters include salt concentration, temperature, and tactile stimulation position; the nematode neural network calculation module is used to generate a stretching signal to the nematode physical information calculation module by taking the changed environment parameters sensed by the C. elegans at the current position as an input during the simulation process of the morphological structure model of the C. elegans in the environment model; the nematode physical information calculation module is used to control the C. elegans to stretch or contract according to the stretching signal and feed back the stretching information of the ventral and dorsal of the C. elegans to the nematode neural network calculation module; the nematode neural network calculation module is further used to continuously generate a stretching signal according to the feedback stretching information and output to the nematode physical information calculation module to realize the interaction with the nematode physical information calculation module; the nematode simulation visualization module is used to visualize the movement form of the C. elegans, the interaction between the C. elegans and the environment, and the neural state of the C. elegans in real time.

Images