CN110989399A - Robot fish bionic control method and system fusing Spiking neural network and CPG - Google Patents
Robot fish bionic control method and system fusing Spiking neural network and CPG Download PDFInfo
- Publication number
- CN110989399A CN110989399A CN201911295113.8A CN201911295113A CN110989399A CN 110989399 A CN110989399 A CN 110989399A CN 201911295113 A CN201911295113 A CN 201911295113A CN 110989399 A CN110989399 A CN 110989399A
- Authority
- CN
- China
- Prior art keywords
- cpg
- model
- neural network
- spiking neural
- excitation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 241000251468 Actinopterygii Species 0.000 title claims abstract description 82
- 238000012421 spiking Methods 0.000 title claims abstract description 70
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 52
- 239000011664 nicotinic acid Substances 0.000 title claims abstract description 41
- 238000000034 method Methods 0.000 title claims abstract description 23
- 230000005284 excitation Effects 0.000 claims abstract description 62
- 238000003062 neural network model Methods 0.000 claims abstract description 20
- 210000002569 neuron Anatomy 0.000 claims description 69
- 230000006870 function Effects 0.000 claims description 35
- 230000003592 biomimetic effect Effects 0.000 claims description 13
- 230000008878 coupling Effects 0.000 claims description 12
- 238000010168 coupling process Methods 0.000 claims description 12
- 238000005859 coupling reaction Methods 0.000 claims description 12
- 230000007613 environmental effect Effects 0.000 claims description 11
- 239000012528 membrane Substances 0.000 claims description 7
- 238000011084 recovery Methods 0.000 claims description 7
- 240000002627 Cordeauxia edulis Species 0.000 claims description 5
- 230000010355 oscillation Effects 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000012905 input function Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 230000033001 locomotion Effects 0.000 description 24
- 210000000225 synapse Anatomy 0.000 description 8
- 238000013178 mathematical model Methods 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000004880 explosion Methods 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 230000027928 long-term synaptic potentiation Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000000946 synaptic effect Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 2
- 230000010365 information processing Effects 0.000 description 2
- 230000005764 inhibitory process Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 description 2
- 210000000006 pectoral fin Anatomy 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009182 swimming Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 230000001020 rhythmical effect Effects 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
- 230000003956 synaptic plasticity Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B17/00—Systems involving the use of models or simulators of said systems
- G05B17/02—Systems involving the use of models or simulators of said systems electric
Abstract
The utility model discloses a robot fish bionic control method and system fusing Spiking neural network and CPG, comprising: the method comprises the steps of establishing a CPG model and a Spiking neural network model, establishing a Spiking neural network and a CPG layered control model, taking the Spiking neural network model as a superior controller, taking the CPG model as a subordinate controller, designing a saturation function to be connected with the superior controller and the subordinate controller, receiving an excitation signal generated by the Spiking neural network model through the saturation function by the CPG model, and outputting a control signal to drive each joint of the bionic robot fish to move.
Description
Technical Field
The disclosure relates to the technical field of motion control, in particular to a robotic fish bionic control method and system fusing a Spiking neural network and a CPG.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
In recent years, people pay more and more attention to abundant marine resources along with the increasing scarcity of land resources. Because the original underwater detection, operation and carrying devices are difficult to meet the requirements of complex underwater operation tasks, the research and development work of the underwater robot is accelerated. The bionic robot fish is used as a combination point of a fish propulsion mechanism and a robot technology, provides a new idea for developing a novel underwater vehicle, and has important research value and application prospect.
With the progress of science and technology, the integration of artificial intelligence and control technology creates a new idea for the bionic robot technology, and particularly, a Central Pattern Generator (CPG) is widely applied to multi-mode swimming control of robotic fish. However, the motion controller based on the CPG has a great disadvantage that the CPG model is designed to simulate the biological motion control, and is mainly used for generating the rhythmic motion signal, the model is difficult to integrate with the environmental information, the environmental adaptability of the robotic fish is problematic,
the Spiking neural network model known as the third generation neural network is the latest research result in the fields of neuroscience and computational intelligence, can simulate the information processing mechanism in real life, and is closer to the actual biological nervous system than the traditional neural network. Spiking neurons have nonlinear processing capabilities for externally input information. Therefore, the Spiking neural network and the CPG are fused to construct the motion control system of the bionic robot fish, and the motion control system has important significance for improving the underwater environment perception level, the swimming autonomy and the mobility of the robot fish.
Disclosure of Invention
In order to solve the problems, the disclosure provides a robot fish bionic control method and system fusing a Spiking neural network and a CPG, an upper-layer decision controller is designed based on the Spiking neural network, a lower-layer motion controller is designed based on the CPG, the problem of difficulty in CPG input design is solved by designing a saturation function, and the Spiking neural network and the CPG are organically fused together to form the robot fish bionic control system. The Spiking neural network simulates human brain neurons to receive external environment information and generate excitation signals, and the excitation signals are input into the CPG neuron model through a saturation function to generate a motion control command so as to drive each joint of the bionic robot fish to move.
In order to achieve the purpose, the following technical scheme is adopted in the disclosure:
the invention provides a robotic fish bionic control method fusing a Spiking neural network and a CPG, which comprises the following steps:
establishing a CPG model: performing dynamic modeling on the robot fish with the pectoral fins at the four joints, determining excitation, downlink and uplink phase coupling coefficients and uplink and downlink coupling coefficient weights at the left side and the right side by using a nonlinear oscillator model as a CPG neuron, and corresponding to the CPG frequency of each joint;
establishing a Spiking neural network model: determining an Izhikevich neuron model, setting parameters to simulate different discharge states, performing neural network training on the different discharge states by adopting an unsupervised algorithm based on an Hebb learning rule, sending the trained data to a CPG neuron as an input signal of the CPG neuron, and driving the CPG to output a control signal of the bionic robotic fish;
establishing a Spiking neural network and CPG layered control model: the Spiking neural network model is used as a superior controller, the CPG model is used as a subordinate controller, a saturation function is designed to be connected with the superior controller and the subordinate controller, the CPG model receives an excitation signal generated by the Spiking neural network model through the saturation function, and a control signal is output to drive each joint of the bionic robot fish to move.
As possible implementation modes, a nonlinear oscillator model is used as a CPG neuron, the input quantity of the CPG neuron is excitation and is divided into left excitation and right excitation, oscillator parameters are obtained through a saturation function, and top-down and bottom-up phase differences are set to respectively drive the left body and the right body of the robot fish.
The Spiking neural network model receives data information of surrounding environment collected by a sensor installed on the bionic robot fish, makes corresponding decision according to environmental change and generates an excitation signal.
As some possible implementations, the determining the Izhikevich neuron model, setting the parameters to simulate different discharge states includes:
an Izhikevich neuron model is used as an upper controller of a layered control model and is expressed as follows;
wherein v is the membrane voltage of the neuron, u is the voltage recovery variable, and I is the external input signal;
the conditions for resetting spiking neurons are:
wherein a, b, c and d are constants, and a represents a time scale parameter of a voltage recovery variable u; b represents the dependence of the neuron on membrane voltage; c and d represent the potential reset values after u and v reach a peak value, respectively.
As some possible implementation manners, the saturation function is used as an input function of a lower-level controller of the CPG model and is divided into a left part and a right part, after an output excitation signal of a Spiking neural network model of the upper-level controller passes through the saturation function, the CPG model of the lower-level controller obtains an oscillator parameter fiAnd AiAnd is used for driving the left and right bodies of the bionic robot fish.
As some possible implementations, the saturation function is expressed as follows:
wherein k is1,k2,b1,b2Represents the constant coefficient, dl,drRespectively representing left and right side excitations, dlowRepresenting the excitation minimum.
As some possible implementations, f in the saturation function1Tail fin and flexible body for bionic machine fish back, saturation function f2Pectoral fins for biomimetic robotic fish when excitation stimulus is less than minimum excitationWhen the excitation signal is gradually enhanced, the oscillation frequency is increased, the amplitude is gradually increased, and the oscillator gradually oscillates; when the excitation signal exceeds the maximum value, the oscillator will not stop and will continue to drive the biomimetic robotic fish with the highest frequency and highest amplitude.
As some possible implementation manners, in the Spiking neural network and the CPG hierarchical control model, the superior controller simulates different discharge states of neurons by adjusting values of parameters a, b, c and d, simulates environmental changes to generate excitation signals, and selects the excitation signals with different discharge states with different characteristics to input the excitation signals into the CPG model through a saturation function;
the CPG model initializes the oscillation frequency and amplitude, outputs a control signal according to the received excitation signal and drives each joint of the bionic robot fish to move.
Compared with the prior art, the beneficial effect of this disclosure is:
the disclosed impulse neural network (SNN-Spiking Neuron Networks) can be used for responding to emergencies caused by various environmental changes by receiving the feedback pertinence of environmental information to simulate twenty-multiple discharge behaviors of biological neurons, so that the underwater complex environment can be better adapted, and the autonomy and the adaptability of a robotic fish system are improved.
According to the robot fish motion system, the appropriate CPG model is established according to the robot fish model, different control commands can be output to drive the robot fish to move by receiving the excitation signal output by the Spiking neural network of the superior controller, and the robot fish motion system has important research significance for researching the bionic robot motion system with hierarchical control.
The layered control system based on the Spiking neural network and the CPG simulates a biological motion mechanism, can be well applied to the motion control of the robot fish, and has great potential for controlling other multi-joint or multi-degree-of-freedom biological robots.
The bionic robot fish motion control system solves the design problem of a complete bionic control system, integrates information input and coding, information processing, high-level decision, motion command transmission, rhythm signal generation, motion signal decoding and output and the like, is just like the whole process of human or higher organisms for processing environmental information, deciding and even generating motion, and a layered control framework provides a complete self-adaptive motion control solution for the bionic robot fish.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and are not to limit the disclosure.
FIG. 1 is a schematic view of a robotic fish biomimetic control system designed in accordance with the present disclosure;
FIG. 2 is a graph of the discharge behavior of 20 properties of a simulated biological neuron according to the present disclosure;
FIG. 3 is a diagram of the CPG control architecture of the present disclosure;
FIG. 4 is a flowchart of the CPG mathematical model control of the present disclosure;
fig. 5 is a Spiking neural network driven CPG flowchart of the present disclosure.
The specific implementation mode is as follows:
the present disclosure is further described with reference to the following drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
Example 1
The embodiment discloses a robot fish bionic control method fusing a Spiking neural network and a CPG (compact peripheral group), wherein a biological motion mechanism is simulated to design a motion control system of a robot fish, and the Spiking neural network is used as an upper-layer controller to process environmental information and generate a decision command; the saturation function enables the CPG model to have an input function; the CPG neuron is used as a subordinate controller to receive spiking excitation signals and output control commands, and the probability that the robotic fish senses the environment underwater and moves autonomously is provided.
Step 1: spiking neural network modeling
The neurons are the basic structural components of the brain, are also the most basic units of the Spiking neural network, and mainly process pulse signals. The Izhikevich neuron model is adopted as an upper controller of a hierarchical control system, and the Izhikevich neuron model has enough functions to form a CPG network and is expressed as follows:
wherein v is the membrane voltage of the neuron, u is the voltage recovery variable, and I is the external input signal.
The conditions for resetting spiking neurons are given by:
wherein a, b, c and d in the formulas (1) and (2) are constants. Wherein the four parameters have the following meanings: a represents a time scale parameter of the recovery variable u; represents the dependence of neurons on membrane voltage; c and d represent the potential reset values after u and v reach a peak value, respectively.
After the spike reaches its peak 30mv, the membrane voltage and recovery variables are reset according to equation (2). If v exceeds 30mv, it is first reset to 30mv so that all spikes have equal amplitude. The model can show all known neuron patterns by changing the parameters a, b, c, d, +30mv is not a threshold but a spike of a pulsed neuron. The neuron model has thresholds of 70mv and 50mv and is dynamic. Fig. 2 simulates the discharge behavior of 20 properties of a biological neuron.
An unsupervised learning method is adopted, and an unsupervised algorithm based on a Hebb learning rule is provided. Synaptic connections between neurons are the neural basis for learning and memory. The connection strength of synapses depends on the degree of stimulation of neurons acting at both ends of the synapse, and this change in synaptic connection strength is called synaptic plasticity. The exact form of STDP varies from synapse to synapse, and can be divided into Long Term Potentiation (LTP), increasing synaptic weight, and long term inhibition (LTD) according to the order of neurons before and after learning.
Wherein A is taken+＝1,A-The STDP has a waveform shown in fig. 3, where τ is 8 ms. LTP and LDP are not symmetrical in time, i.e. t>When 0, the synapse plasticity is enhanced for a long time, and the synapse weight of two continuous neurons is increased; on the contrary, when t is<At 0, the synapse plasticity is inhibited for a long time, the synapse weight value between two connected neurons becomes smaller, and the asymmetry can prevent unstable neuron transmission signals caused by the increase or inhibition.
The robotic fish used in the present disclosure has four joints and a pair of pectoral fins, and can perform a reciprocating oscillating motion. The CPG control structure of the robotic fish is shown in FIG. 4.
The CPG signal is used for driving the swing of each joint, and a phase oscillator model is adopted to express the following formula (4):
in the formula [ theta ]iAnd riRepresenting the phase and amplitude respectively for the state variable of the oscillator; f. ofiAnd RiDetermining the internal frequency and amplitude of the oscillator; tau isiFor normal amount, r is determinediConverge to RiThe speed of (d); the mutual coupling relationship between oscillators is defined by weight wijPhase difference of sum
It is determined that i represents the number of CPGs used by the hierarchical control system of the present disclosure, and the value of i is 1.
In the embodiment, the robotic fish is controlled by the steering engine, and the action of the steering engine is driven by the output signal of the oscillator; a nonlinear oscillator model is established for the robot fish with four joints and a pair of pectoral fins, and the straight-swimming and turning motions of the robot fish are controlled.
The nonlinear oscillator model is used as a CPG neuron, the input quantity of the CPG neuron is excitation, and the CPG neuron is divided into left excitation dlAnd right excitation drObtaining the oscillator parameter f after the saturation functioniAnd AiDriving the left and right body, respectively; behavior phi under phase coupling relation of definition modeldUpward is phiu。
The method determines various parameters of the robot fish during straight-swimming and turning motion, such as an uplink phase coupling coefficient, a downlink phase coupling coefficient, tail fin extension length, CPG frequency left-right excitation values corresponding to various joints and the like, and a CPG mathematical model control flow chart is shown in fig. 5.
Step 2.1: defining CPG model key parameters and carrying out robot fish control simulation
The present disclosure employs MATLAB software to perform simulations on this platform. The driving signal d is set to drive two sides of the robot fish, the driving signal d is increased from 0 to 5, other parameters of the CPG model are shown in table 1, joints and tail fin neurons of the robot fish oscillate in sequence, the frequency and the amplitude of the neurons are increased along with the increase of the driving signal, and the control signal of each joint servo motor is obtained, and each shutdown is controlled by two neurons with the phase difference of 180 degrees.
TABLE 1 CPG god-element mathematical model parameters
| Parameter(s) | |
|
Joint 3 | Tail fin | |
| n | |||||
| 2 | 2 | 2 | 2 | 2 | |
| τ | 20 | 20 | 20 | 20 | 20 |
| dl | 2.5 | 2 | 1.5 | 1 | 1 |
| dr | 5 | 5 | 5 | 5 | 5 |
| k1 | 0.45 | 0.45 | 0.45 | 0.45 |
Step 3, designing a saturation function
Designing a saturation function as an input function of the CPG lower-level controller, wherein the saturation function is divided into a left part and a right part, the CPG lower-level controller obtains an oscillator parameter after an output excitation signal of the Spiking upper-level controller passes through the saturation function, the CPG lower-level controller has two input signals for driving one of the left side and the right side of the bionic robot fish, and the saturation function can be expressed as follows:
wherein k is1,k2,b1,b2Represents the constant coefficient, dl,drRespectively representing the left and right side excitation, dlowRepresenting the excitation minimum.
(5) The saturation function in the formula is used for the tail fin and the flexible body on the back of the machine fish, and the saturation function in (6) is used for the pectoral fin of the machine fish.
When the excitation stimulus is less than the minimum excitation, the oscillator stops oscillating. With the gradual enhancement of the excitation signal, the oscillation frequency is increased, and the amplitude is gradually increased; the oscillator oscillates gradually. When the excitation signal exceeds the maximum value, the oscillator will not stop and will continue to drive the biomimetic robotic fish with the highest frequency and highest amplitude.
Step four, designing basic control flow chart of CPG unit driven by Spiking neural network
Through the steps, the modeling design of each level of controller of the hierarchical control system is completed, the superior controller adopts an Izhikevich neuron model, different discharge states of rule explosion, internal explosion, fast reading and issuing and the like of the neuron can be simulated by adjusting the values of the parameters a, b, c and d, the discharge states with different characteristics are selected as the input of the CPG mathematical model according to the characteristic, and the basic control flow chart of the Spiking neural network driving the CPG unit is shown in figure 5.
The Spiking neural network is used as an upper-level controller of a layered structure of the bionic control system, processes sensor and environment information, makes a decision on environment change and generates an excitation signal;
the CPG is a lower-level controller of a layered structure of the bionic control system, receives an excitation signal generated by a Spiking neural network, and outputs a control command to drive each joint of the bionic robot fish to move;
the saturation function is used for connection interaction between the upper-level controller and the lower-level controller, so that the CPG has an enhanced input interface and can receive an excitation signal output by the Spiking neural network upper-level controller;
determining a control flow of driving the CPG neuron by the Spiking neural network: based on the adopted Izhikevich neuron model, the discharge states with different characteristics such as regular explosion, internal explosion, fast reading and releasing and the like of the simulation neuron simulate environmental changes by adjusting key parameters, and the discharge states with different characteristics are selected as the input of the CPG model. And simulating the motion control of the robot fish in the matlab environment to obtain an excitation signal output by the Spiking neural network and a control signal of each joint of the robot fish.
Spiking neural network driven CPG control simulation
The method adopts MATLAB simulation software to set a group of Spiking neural network Izhikevich neuron model key parameters a, b, c and d as external excitation signals to drive the CPG mathematical model of the bionic robot fish, and simulates the straight-swimming state and the turning state of the bionic robot fish. Parameter setting as shown in table 2, the state parameters of the Spiking neural network of the superior controller and the excitation signal obtained as output are set, 0/1 pulse sequences can be obtained by the Spiking neural network and input to the CPG unit of the inferior controller as the driving signal d, the output result of the CPG unit after the pulse sequence driving signal is input can be obtained, and the control signal of all joints of the robotic fish when the Spiking neural network is driven can be generated.
TABLE 2 Spiking neural network parameters
| Parameter(s) | a | b | c | d | I |
| Set value | 1.0 | 1.5 | -60 | 0 | -65 |
Example 2
The utility model provides a machine fish bionic control system that fuses Spiking neural network and CPG includes:
the CPG model establishing module is used for performing dynamic modeling on the robot fish with the pectoral fins at the four joints, determining excitation, downlink and uplink phase coupling coefficients and uplink and downlink coupling coefficient weights at the left side and the right side by using a nonlinear oscillator model as CPG neurons, and corresponding to the CPG frequency of each joint;
the Spiking neural network model establishing module is used for determining an Izhikevich neuron model, setting parameters to simulate different discharge states, performing neural network training on the different discharge states by adopting an unsupervised algorithm based on a Hebb learning rule, sending the trained data to the CPG neuron, and driving the CPG neuron to output a bionic robot fish control signal as an input signal of the CPG neuron;
the Spiking neural network and CPG layered control model building module is used for enabling a piking neural network model to serve as a superior controller, enabling a CPG model to serve as a subordinate controller, designing a saturation function to be connected with the superior controller and the subordinate controller, receiving an excitation signal generated by the Spiking neural network model through the saturation function by the CPG model, and outputting a control signal to drive each joint of the bionic robot fish to move.
The saturation function is expressed as follows:
wherein k is1,k2,b1,b2Represents the constant coefficient, dl,drRespectively representing left and right side excitations, dlowRepresenting the excitation minimum.
The above is merely a preferred embodiment of the present disclosure and is not intended to limit the present disclosure, which may be variously modified and varied by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.
Although the present disclosure has been described with reference to specific embodiments, it should be understood that the scope of the present disclosure is not limited thereto, and those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present disclosure.
Claims (10)
1. A robotic fish bionic control method fusing a Spiking neural network and a CPG is characterized by comprising the following steps:
establishing a CPG model: performing dynamic modeling on the robot fish with the pectoral fins at the four joints, determining excitation, downlink and uplink phase coupling coefficients and uplink and downlink coupling coefficient weights at the left side and the right side by using a nonlinear oscillator model as a CPG neuron, and corresponding to the CPG frequency of each joint;
establishing a Spiking neural network model: determining an Izhikevich neuron model, setting parameters to simulate different discharge states, performing neural network training on the different discharge states by adopting an unsupervised algorithm based on an Hebb learning rule, sending the trained data to a CPG neuron as an input signal of the CPG neuron, and driving the CPG to output a control signal of the bionic robotic fish;
establishing a Spiking neural network and CPG layered control model: the Spiking neural network model is used as a superior controller, the CPG model is used as a subordinate controller, a saturation function is designed to be connected with the superior controller and the subordinate controller, the CPG model receives an excitation signal generated by the Spiking neural network model through the saturation function, and a control signal is output to drive each joint of the bionic robot fish to move.
2. The method for controlling the biomimetic of the robotic fish fusing the Spiking neural network and the CPG according to claim 1,
a nonlinear oscillator model is used as a CPG neuron, the input quantity of the CPG neuron is excitation and is divided into left excitation and right excitation, oscillator parameters are obtained through a saturation function, and the left body and the right body of the robot fish are respectively driven by setting phase differences from top to bottom and from bottom to top.
3. The method for controlling the biomimetic of the robotic fish fusing the Spiking neural network and the CPG according to claim 1,
the Spiking neural network model receives data information of surrounding environment collected by a sensor installed on the bionic robot fish, makes corresponding decision according to environmental change and generates an excitation signal.
4. The method for controlling biomimetic robotic fish fused with Spiking neural network and CPG according to claim 1, wherein the determining Izhikevich neuron model, setting parameters to simulate different discharge states comprises:
an Izhikevich neuron model is used as an upper controller of a layered control model and is expressed as follows;
wherein v is the membrane voltage of the neuron, u is the voltage recovery variable, and I is the external input signal;
the conditions for resetting spiking neurons are:
wherein a, b, c and d are constants, and a represents a time scale parameter of a voltage recovery variable u; b represents the dependence of the neuron on membrane voltage; c and d represent the potential reset values after u and v reach a peak value, respectively.
5. The method as claimed in claim 1, wherein the saturation function is divided into left and right parts as the input function of the lower controller of the CPG model, and the CPG model of the lower controller obtains the oscillator parameter f after the output excitation signal of the upper controller of the Spiking neural network model passes through the saturation functioniAnd AiFor driving the left and right bodies of the biomimetic robotic fish。
6. The method for controlling biomimetic robotic fish fused with Spiking neural network and CPG according to claim 5, wherein the saturation function is expressed as follows:
wherein k is1,k2,b1,b2Represents the constant coefficient, dl,drRespectively representing left and right side excitations, dlowRepresenting the excitation minimum.
7. The method as claimed in claim 6, wherein f in the saturation function is f, and f is a function of the saturation function1Tail fin and flexible body for bionic machine fish back, saturation function f2Pectoral fins for biomimetic robotic fish;
when the excitation stimulus is smaller than the minimum excitation, the oscillator stops oscillating, the oscillation frequency increases along with the gradual enhancement of the excitation signal, the amplitude gradually increases, and the oscillator gradually oscillates; when the excitation signal exceeds the maximum value, the oscillator will not stop and will continue to drive the biomimetic robotic fish with the highest frequency and highest amplitude.
8. The method for controlling the biomimetic of the robotic fish fusing the Spiking neural network and the CPG according to claim 1,
the superior controller simulates different discharge states of neurons by adjusting values of parameters a, b, c and d, simulates environmental changes to generate excitation signals, and selects the excitation signals with different characteristic discharge states to input into the CPG model through a saturation function;
the CPG model initializes the oscillation frequency and amplitude, outputs a control signal according to the received excitation signal and drives each joint of the bionic robot fish to move.
9. A robotic fish bionic control system fusing a Spiking neural network and a CPG is characterized by comprising:
the CPG model establishing module is used for performing dynamic modeling on the robot fish with the pectoral fins at the four joints, determining excitation, downlink and uplink phase coupling coefficients and uplink and downlink coupling coefficient weights at the left side and the right side by using a nonlinear oscillator model as CPG neurons, and corresponding to the CPG frequency of each joint;
the Spiking neural network model establishing module is used for determining an Izhikevich neuron model, setting parameters to simulate different discharge states, performing neural network training on the different discharge states by adopting an unsupervised algorithm based on a Hebb learning rule, sending the trained data to the CPG neuron, and driving the CPG neuron to output a bionic robot fish control signal as an input signal of the CPG neuron;
the Spiking neural network and CPG layered control model building module is used for enabling a piking neural network model to serve as a superior controller, enabling a CPG model to serve as a subordinate controller, designing a saturation function to be connected with the superior controller and the subordinate controller, receiving an excitation signal generated by the Spiking neural network model through the saturation function by the CPG model, and outputting a control signal to drive each joint of the bionic robot fish to move.
10. The robotic fish biomimetic control system fusing the Spiking neural network and the CPG according to claim 9, wherein the saturation function is expressed as follows:
wherein k is1,k2,b1,b2Represents the constant coefficient, dl,drRespectively representing left and right side excitations, dlowRepresenting the excitation minimum.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911295113.8A CN110989399A (en) | 2019-12-16 | 2019-12-16 | Robot fish bionic control method and system fusing Spiking neural network and CPG |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911295113.8A CN110989399A (en) | 2019-12-16 | 2019-12-16 | Robot fish bionic control method and system fusing Spiking neural network and CPG |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110989399A true CN110989399A (en) | 2020-04-10 |
Family
ID=70094217
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911295113.8A Pending CN110989399A (en) | 2019-12-16 | 2019-12-16 | Robot fish bionic control method and system fusing Spiking neural network and CPG |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110989399A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112904873A (en) * | 2021-01-26 | 2021-06-04 | 西湖大学 | Bionic robot fish control method and device based on deep reinforcement learning and storage medium |
| CN113433865A (en) * | 2021-07-09 | 2021-09-24 | 西安交通大学 | Design method and system of IPMC-driven bionic wave fin motion control system |
| CN114460849A (en) * | 2022-04-12 | 2022-05-10 | 北京晟海汇泽科技有限公司 | Bionic robot fish motion control method and device and bionic robot fish |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101776863A (en) * | 2009-11-17 | 2010-07-14 | 中国科学院自动化研究所 | Movement control method of pectoral fin impelling type machine fish |
| CN104050506A (en) * | 2014-06-24 | 2014-09-17 | 电子科技大学 | Aircraft conflict detection method based on Spiking neural network |
| CN110286592A (en) * | 2019-06-28 | 2019-09-27 | 山东建筑大学 | A kind of multi-modal movement technique of machine fish based on BP neural network and system |
| CN110488611A (en) * | 2019-09-02 | 2019-11-22 | 山东建筑大学 | A kind of biomimetic robot fish movement control method, controller and bionic machine fish |
-
2019
- 2019-12-16 CN CN201911295113.8A patent/CN110989399A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101776863A (en) * | 2009-11-17 | 2010-07-14 | 中国科学院自动化研究所 | Movement control method of pectoral fin impelling type machine fish |
| CN104050506A (en) * | 2014-06-24 | 2014-09-17 | 电子科技大学 | Aircraft conflict detection method based on Spiking neural network |
| CN110286592A (en) * | 2019-06-28 | 2019-09-27 | 山东建筑大学 | A kind of multi-modal movement technique of machine fish based on BP neural network and system |
| CN110488611A (en) * | 2019-09-02 | 2019-11-22 | 山东建筑大学 | A kind of biomimetic robot fish movement control method, controller and bionic machine fish |
Non-Patent Citations (1)
| Title |
|---|
| 李旭: "基于CPG的仿生机器鱼运动控制优化", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112904873A (en) * | 2021-01-26 | 2021-06-04 | 西湖大学 | Bionic robot fish control method and device based on deep reinforcement learning and storage medium |
| CN112904873B (en) * | 2021-01-26 | 2022-08-26 | 西湖大学 | Bionic robot fish control method and device based on deep reinforcement learning |
| CN113433865A (en) * | 2021-07-09 | 2021-09-24 | 西安交通大学 | Design method and system of IPMC-driven bionic wave fin motion control system |
| CN114460849A (en) * | 2022-04-12 | 2022-05-10 | 北京晟海汇泽科技有限公司 | Bionic robot fish motion control method and device and bionic robot fish |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pathak et al. | Learning to control self-assembling morphologies: a study of generalization via modularity | |
| Yu et al. | A survey on CPG-inspired control models and system implementation | |
| Lewis et al. | Genetic algorithms for gait synthesis in a hexapod robot | |
| CN110989399A (en) | Robot fish bionic control method and system fusing Spiking neural network and CPG | |
| Arena et al. | An adaptive, self-organizing dynamical system for hierarchical control of bio-inspired locomotion | |
| Bal et al. | CPG-based autonomous swimming control for multi-tasks of a biomimetic robotic fish | |
| Torresen | A scalable approach to evolvable hardware | |
| Guerra-Hernandez et al. | A FPGA-based neuromorphic locomotion system for multi-legged robots | |
| Zhao et al. | Development of multi-mode biomimetic robotic fish based on central pattern generator | |
| Lu et al. | Serpentine locomotion of a snake-like robot controlled by cyclic inhibitory CPG model | |
| Hu et al. | Learning a faster locomotion gait for a quadruped robot with model-free deep reinforcement learning | |
| Zhao et al. | Design and CPG-based control of biomimetic robotic fish | |
| Zhang et al. | From simulation to reality: A learning framework for fish-like robots to perform control tasks | |
| Wang et al. | A CPG-based sensory feedback control method for robotic fish locomotion | |
| Na et al. | Locomotion generator for robotic fish using an evolutionary optimized central pattern generator | |
| Chen et al. | A bio-inspired spiking neural network for control of a 4-dof robotic arm | |
| Wen et al. | Cpg control model of snake-like robot parameters of optimization based on ga | |
| Markovikj | Deep apprenticeship learning for playing games | |
| Patel et al. | Super-lampreys and wave energy: Optimised control of artificially-evolved, simulated swimming lamprey | |
| Wang et al. | Locomotion control of robotic fish with a hierarchical framework combining spiking neural networks and CPGs | |
| Zharinov et al. | CPG as a controller for biomimetic floating robots | |
| Bing | Biological-Inspired Hierarchical Control of a Snake-like Robot for Autonomous Locomotion | |
| Bahramianl et al. | Introducing neural-network based model and pretraining method to design central pattem generator (CPG) | |
| Scola et al. | A neuromorphic control architecture inspired by the limbic system | |
| Foong et al. | Generating oscillation activity with echo state network to mimic the behavior of a simple central pattern generator |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200410 |