A CPG control system for a quadruped robot and its parameter setting method

Through the time-delay coupled oscillator unit and nonlinear dynamic analysis using a one-way ring network structure in a four-legged robot, the control and parameter setting of multiple gait modes are realized, and the system complexity and resilience in the prior art are solved. The awkward problem improves the regulation performance and system stability of the gait mode.

CN115202259BActive Publication Date: 2024-11-08TONGJI UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210866251.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-22
Publication Date
2024-11-08
Estimated Expiration
2042-07-22

AI Technical Summary

Technical Problem

The existing four-legged robot CPG control system has a complex structure, high computational complexity, and lacks a unified parameter setting method, resulting in poor system unadjustability and robustness.

Method used

Four oscillator units with a one-way ring network structure are used for time-delay coupling, and the control of multiple gait modes is achieved through nonlinear dynamic analysis, including four-leg jumping, lateral sequential walking, bipedal jumping, diagonal sequential walking, Pacing and diagonal trotting, using Hopf bifurcation theory and numerical methods for parameter tuning.

Benefits of technology

The gait control system of four-legged robot with a simple structure is realized, which improves the gait mode regulation performance and the robustness of the control system. The gait mode switching and stability are realized through a single time delay parameter, reducing the complexity of the engineering implementation. sex.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115202259B_ABST
    Figure CN115202259B_ABST
Patent Text Reader

Abstract

The present invention relates to a CPG control system of a quadruped robot and a parameter setting method thereof. The system comprises four oscillator units corresponding to the hip joints of the quadruped robot. The oscillator units are connected to form a unidirectional ring network structure, and the connection mode is time-delay coupling. Under the action of time-delay, the oscillator units realize multiple stable phase differences based on a spontaneous symmetry breaking mechanism. Through nonlinear dynamic analysis, the parameters of the CPG control system are set, thereby realizing multiple gait modes of the quadruped robot and their switching. The multiple gait modes include four-legged hopping, lateral sequential walking, two-legged hopping, diagonal sequential walking, pacing and diagonal trotting. Compared with the prior art, the system structure of the present invention is simple and easy to implement in engineering; the parameter setting method is reliable, and the change of the gait mode can be realized by switching a single time-delay parameter between rhythmic regions; within a given rhythmic region, the gait mode will not change with the change of the parameter, and the robustness is effectively guaranteed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of bionic robots, in particular to a CPG control system of a quadruped robot and a parameter setting method thereof. Background Art

[0002] The research of bionic control is of great significance to the progress of robotics. In recent years, bionic control simulating biological central pattern generators (CPG) has become one of the research hotspots in the field of bionic control. CPG can spontaneously generate stable rhythmic movements in the absence of high-level control signals and external feedback. This movement has the advantages of being easily regulated by high-level commands, strong coupling, strong adaptability, and simple structure. It can reduce the workload of the control system and save working time.

[0003] With the development of quadruped robots and the rise of artificial intelligence in recent years, robot autonomy and intelligence have become an important part of quadruped robot research. Quadruped robots have gradually acquired the ability to perceive the environment, autonomously plan and interact with the environment. Researchers are also more focused on improving the autonomous adaptability and functionality of robots. At present, most quadruped robots can achieve various motion gaits and complete simple tasks through control algorithms. In the field of quadruped robot gait control, CPG is one of the simplest control methods. By constructing a suitable rhythm controller, each leg of the quadruped robot is assigned a controller so that the robot can move according to a certain timing. Although this method can realize the gait movement of the quadruped robot, it still has the following major shortcomings: (1) The proposed CPG structural unit must be an oscillator model of two dimensions or above, and the network structure mostly adopts a fully connected four-oscillator unit, which leads to a high dimension of the entire CPG system and increased computational complexity; (2) There is no general and unified dynamic analysis method for parameter setting. It can only be based on the specific motion mode of the robot, using trial and error or optimization calculation methods to give one or several sets of fixed parameter values. The actual engineering implementation is relatively complicated, and the adjustability of the system and the robustness of the control system cannot be guaranteed. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a CPG control system of a quadruped robot and a parameter setting method thereof, which can realize a control system with a simple structure and improve the control performance of the gait pattern and the robustness of the control system.

[0005] The objective of the present invention can be achieved through the following technical scheme: a CPG control system of a quadruped robot, comprising four oscillator units for correspondingly controlling the hip joints of the quadruped robot, wherein the four oscillator units are connected to form a unidirectional ring network structure, and the connection mode between the oscillator units is time-delay coupling. Under the action of time delay, the oscillator units realize a variety of stable phase differences based on a spontaneous symmetry breaking mechanism, and the parameters of the CPG control system are adjusted through nonlinear dynamic analysis, thereby realizing a variety of gait modes of the quadruped robot and their switching, wherein the various gait modes include quadrupedal hopping (Pronk), lateral walking (LS walk), bipedal hopping (Bound), diagonal walking (D-Swalk), pacing (Pace) and diagonal trotting (Trot).

[0006] Furthermore, the unidirectional ring network structure is specifically a single ring network to realize four gait modes, and the four gait modes are divided into three cases:

[0007] a. quadrupedal hopping, lateral walking, diagonal walking and bipedal hopping;

[0008] b. quadrupedal hopping, lateral walking, diagonal walking and pacing;

[0009] c. Four-legged hopping, sideways walking, diagonal walking and diagonal trotting.

[0010] Furthermore, the unidirectional ring network structure is specifically a superposition of three single ring networks to realize all six gait modes.

[0011] Furthermore, the oscillator units are connected using a nonlinear time-delay coupling function or a linear time-delay coupling function.

[0012] Furthermore, the oscillator unit adopts an oscillator model with a periodic rhythm signal.

[0013] Furthermore, the oscillator unit adopts a node model without a periodic rhythm signal.

[0014] Furthermore, the models and coupling functions adopted by the four oscillator units are the same, so that the CPG control system has symmetry.

[0015] Furthermore, the nonlinear dynamic analysis is specifically a Hopf bifurcation analysis. Based on the Hopf bifurcation theory, the CPG control system can produce periodic rhythmic stability switching under the action of time lag.

[0016] Furthermore, the parameter region corresponding to the periodic rhythm can produce a stable phase difference based on the symmetry breaking mechanism, which is the parameter region of the gait pattern.

[0017] A method for setting parameters of a CPG control system of a quadruped robot comprises the following steps:

[0018] S1. Linearize the time-delay coupled nonlinear system near the equilibrium point to obtain the corresponding linear system and the corresponding characteristic equation;

[0019] S2. Assume that the characteristic equation has a pair of pure imaginary roots with zero real part, substitute the assumed pure imaginary roots into the characteristic equation, separate the real part and the imaginary part, and obtain two equations of the real part and the imaginary part. Then, according to the relationship between trigonometric functions, obtain the algebraic equation corresponding to the periodic rhythm;

[0020] S3. Calculate the algebraic equation to get the corresponding frequency value, substitute it into the real part and imaginary part equations to calculate the critical lag value;

[0021] S4. Verify the parameter plane determined by the critical time lag. Take any value in the region and use numerical methods to verify that the region is a rhythmic mode or a resting mode. The rhythmic mode region and the resting mode region appear alternately. The corresponding rhythmic mode region is distributed with four-legged jumping (Pronk), lateral walking (LS walk), two-legged jumping (Bound), and diagonal walking (DS walk) modes in sequence.

[0022] By adjusting the network topology, the parameter region of the bipedal hopping (Bound) can further realize the pacing (Pace) and diagonal trotting (Trot) modes.

[0023] Compared with the prior art, the present invention has the following advantages:

[0024] 1. The present invention aims at the phase relationship in the gait mode of a quadruped robot, realizes multiple stable phase differences based on the spontaneous symmetry breaking mechanism, and proposes a unidirectional ring-delay coupled CPG quadruped robot gait control system, which includes four oscillator units and a unidirectional ring network topology structure between the units. The network structure is simple, and each oscillator can be expressed by a node model with only one variable, which is easy to implement in engineering. By determining the parameter range under different gait modes through nonlinear dynamic analysis, the quadruped robot can realize multiple gait modes and their switching.

[0025] 2. For the CPG gait control system proposed in the present invention, the present invention also proposes a general and unified dynamic analysis method for parameter setting. The parameter setting method is simple and easy to implement. The nonlinear dynamic analysis of the control system can give the parameter areas corresponding to different gait patterns. The gait pattern can be easily regulated. A single time-delay parameter can be switched between rhythm areas to achieve the change of the gait pattern. There is only one gait pattern in a parameter area. Within a given rhythm area, the gait pattern will not change with the change of parameters, so that the control system has good robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1a This is a schematic diagram of the first unidirectional ring network structure of the quadruped robot CPG of the present invention;

[0027] Figure 1b This is a schematic diagram of a second unidirectional ring network structure of the quadruped robot CPG of the present invention;

[0028] Figure 1c Schematic diagram of the third unidirectional ring network structure of the quadruped robot CPG of the present invention;

[0029] Figure 1d Schematic diagram of three unidirectional ring network superposition structures of the quadruped robot CPG of the present invention;

[0030] Figure 2 It is the parameter area corresponding to different step modes in the embodiment;

[0031] Figure 3 Schematic diagram of CPG control signal output when the quadruped robot is in a pronk gait mode in the embodiment;

[0032] Figure 4 Schematic diagram of CPG control signal output when the quadruped robot is in LS walk gait mode in the embodiment;

[0033] Figure 5 Schematic diagram of CPG control signal output when the quadruped robot is in a bipedal hopping (Bound) gait mode in the embodiment;

[0034] Figure 6 Schematic diagram of CPG control signal output when the quadruped robot is in diagonal sequential walking (DS walk) gait mode in the embodiment;

[0035] Figure 7 Schematic diagram of CPG control signal output when the quadruped robot is in pacing mode in the embodiment;

[0036] Figure 8Schematic diagram of CPG control signal output when the quadruped robot is in the diagonal trot (Trot) gait mode in the embodiment. DETAILED DESCRIPTION

[0037] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Example

[0039] like Figure 1a to Figure 1d As shown, a CPG control system of a quadruped robot includes four oscillator units (as shown in the figure Osc.1, Osc.2, Osc.3, Osc.4), and the four oscillator units are connected to form a unidirectional ring network topology structure, and the connection method is time-delay coupling. The unidirectional ring network structure adopted can be a single ring network (as shown in the figure Figure 1a to Figure 1c As shown in the figure, four gait modes can be realized at this time: one of bipedal hopping (Bound), pacing (Pace) and diagonal trot (Trot), as well as quadrupedal hopping (Pronk), lateral sequence walking (LS walk), and diagonal sequence walking (DS walk);

[0040] It can also be a superposition of three single ring networks (such as Figure 1d As shown in the figure, all six gait modes of the quadruped robot can be realized - four-legged hopping (Pronk), lateral walking (LS walk), two-legged hopping (Bound), diagonal walking (D-Swalk), pacing (Pace) and diagonal trot (Trot).

[0041] In practical applications, the four oscillator units that control the hip joints of the quadruped can be oscillator models with periodic rhythmic signals or node models without periodic rhythmic signals. In addition, the connection method adopted between the oscillator units can be a linear function or a nonlinear function, but the oscillator unit signals must be delayed time-delay coupling when interacting with each other, and the time lag is a control parameter used to control the gait pattern. Each oscillator unit can use exactly the same mathematical model or a different mathematical model. In this embodiment, the same oscillator unit model and coupling function are selected. Under the unidirectional ring structure network topology, the constructed CPG gait control system has symmetry.

[0042] After constructing the CPG control system, the parameter region in which the output signal of each oscillator unit has a periodic rhythm is determined through nonlinear dynamic analysis to achieve parameter setting of the CPG network system. By utilizing the change of the time-delay parameter between parameter regions, the multiple gait modes of the quadruped robot and their switching can be realized.

[0043] In this embodiment, parameter setting is specifically determined by bifurcation analysis of nonlinear dynamics. Based on the Hopf bifurcation theory, the CPG control system can produce stability switching of periodic rhythm under the action of time lag, and the parameter region corresponding to the periodic rhythm is the parameter region of the gait pattern.

[0044] The specific steps of parameter tuning are as follows: firstly, linearize the time-delay coupled nonlinear system near the equilibrium point to obtain the corresponding linear system and give the corresponding characteristic equation;

[0045] In order to find the parameter conditions that can produce periodic rhythms, it is assumed that the characteristic equation has a pair of pure imaginary roots with zero real part; the assumed pure imaginary roots are substituted into the characteristic equation, and the real part and the imaginary part are separated to obtain two equations of the real part and the imaginary part. According to the relationship between trigonometric functions, the algebraic equation corresponding to the periodic rhythm is obtained;

[0046] Calculate the algebraic equation to get the corresponding frequency value, substitute it back into the real and imaginary part equations, and calculate the critical lag value;

[0047] The parameter plane determined by the critical time lag is examined. Any value is taken in the area, and the numerical method is used to verify that the area is a rhythmic mode or a resting state mode. The rhythmic mode area and the resting state mode area appear alternately, and the corresponding rhythmic mode area is distributed with quadrupedal jumping (Pronk), lateral walking (LS walk), bipedal jumping (Bound), and diagonal walking (DS walk) modes in sequence. By adjusting the network topology, the parameter area of ​​bipedal jumping (Bound) can realize pacing (Pace) and diagonal trotting (Trot).

[0048] In this embodiment, a VDP oscillator expressed by a variable is used, and the coupling function adopts a simple linear time-delay coupling. In terms of network topology, a unidirectional ring network topology of 1→2→4→3→1 order is selected, such as Figure 1a As shown in the TypeI structure in the figure, the following quadruped robot CPG control system is constructed:

[0049]

[0050] Among them, the parameters α>0 and β>0 are used to adjust the frequency and amplitude of the periodic rhythm signal, ε>0 is the nonlinear damping coefficient in the VDP system, which is used to control the shape of the periodic rhythm, τ>0 is the coupling delay between the VDP oscillators, k is the coupling connection weight, and x i is the output signal of the i-th VDP oscillator, are the corresponding velocity and acceleration signals respectively.

[0051] Substitute x with a variable i =u i , The equivalent system model is as follows:

[0052]

[0053] For the above-mentioned time-delay coupled CPG control system, the parameter region of the periodic rhythm can be obtained by using the Hopf bifurcation theory and numerical methods of time-delay dynamics. This process is also a general and unified dynamic analysis method for system parameter tuning. Specifically:

[0054] The above nonlinear system is linearized near the equilibrium point to obtain the characteristic equation of its linear system, as follows:

[0055] g1(λ,τ)=β+(k-εα)λ+λ 2 -kλe -λτ =0

[0056] g2(λ,τ)=β+(k-εα)λ+λ 2 +kλe -λτ =0

[0057] g3(λ,τ)=β+(k-εα)λ+λ 2 -ikλe -λτ =0

[0058] g4(λ,τ)=β+(k-εα)λ+λ 2 +ikλe -λτ =0

[0059] Under the action of time delay τ>0, the CPG system can generate a stable periodic rhythm signal. Based on the Hopf bifurcation theory, let λ=iω, separate the real part and the imaginary part to obtain:

[0060]

[0061]

[0062] At this time, the frequency of the periodic rhythm signal should satisfy the following algebraic equation:

[0063] ω 4 +(ε 2 α 2 -2β-2kεα)ω 2 +β 2 =0

[0064] After calculation, the exact solution of the frequency is obtained, which is:

[0065]

[0066] At this time ±The condition of >0 must be met, so the critical value of the lag of the periodic rhythm is calculated as follows:

[0067]

[0068]

[0069] According to the eigenvector corresponding to the eigenroot, the above eigenroot has the following eigenvector:

[0070]

[0071] Among them, the vector η + =(1,iω + ) corresponds to the critical value of the time lag Vector η - =(1,iω - ) corresponds to j=0,1,…,n, that is, the periodic rhythm caused by the time delay satisfies the above-mentioned phase difference, thereby forming a phase relationship of the gait pattern, and a variety of gait patterns are formed in different areas.

[0072] In order to prove that the parameter range adjusted by the above method can meet the gait pattern of the quadruped robot, this embodiment fixes the system parameters α=1,β=1,ε=0.03. Based on the above analytical theoretical method, the parameter range of the periodic rhythm is given in the (τ, k) plane, as follows: Figure 2 shown.

[0073] exist Figure 2 In the figure, AD represents the amplitude dead region of the system. At this time, each oscillator of the CPG system does not generate a periodic rhythm signal. Outside the AD region, the system can generate a periodic rhythm pattern. In particular, according to the eigenvector corresponding to the eigenroot, the Pronk region will generate a completely synchronized periodic rhythm of the four oscillators, which is the four-legged jumping (Pronk) gait pattern of the quadruped robot. The output signal is as follows: Figure 3 shown.

[0074] When the periodic rhythm passes through the AD region, a delayed synchronization with a phase lag of 1 / 4 cycle will occur, corresponding to the LS walk gait pattern of the quadruped robot, such as Figure 4 shown.

[0075] The periodic rhythm corresponding to the bound area has a phase lag synchronization of 1 / 2 cycle. Therefore, by changing the coupling connection direction between units, the bipedal hopping (Bound) gait pattern (such as Figure 5 ), Pace gait pattern (such as Figure 7 ), diagonal trot (Trot) gait pattern (such as Figure 8 ), the corresponding network structures are Type I (corresponding to Figure 1a ), Type II (corresponding to Figure 1b ), Type III (corresponding to Figure 1c ). Therefore, in order to realize the above six gaits, the superposition of three unidirectional ring network structures, Type I, Type II, and Type III, can be used, namely Figure 1d The hybrid network structure in .

[0076] The periodic rhythm oscillation corresponding to the DS walk area has a phase difference of 1 / 4 cycle phase advance synchronization. At this time, the corresponding quadruped robot gait mode is the diagonal sequence walking (DS walk) gait mode, such as Figure 6 shown.

[0077] In summary, the CPG quadruped robot control system proposed in the present technical solution has a simple structure, each oscillator unit can be expressed by only one variable, and is easy to implement in engineering; the present technical solution uses CPG system dynamics analysis to perform parameter tuning. Compared with the traditional trial and error method, the parameter tuning method is simple, and the parameter areas under different gait modes can be given through nonlinear dynamics analysis of the control system; the gait mode is simple to control, and the switching of a single time-delay parameter between rhythm areas can achieve the change of gait mode; within a given rhythm area, the gait pattern will not change with the change of parameters, and the constructed control system has good robustness.

Claims

1. A parameter setting method for CPG control system of quadruped robot, It is characterized in that The CPG control system of the quadruped robot includes four oscillator units for correspondingly controlling the hip joints of the quadruped robot. The four oscillator units are connected to form a unidirectional ring network structure. The connection mode between the oscillator units is time-delay coupling. Under the action of time delay, the oscillator units realize multiple stable phase differences based on the spontaneous symmetry breaking mechanism. Through nonlinear dynamic analysis, the parameters of the CPG control system are set, thereby realizing multiple gait modes of the quadruped robot and their switching. The multiple gait modes include four-legged hopping, lateral sequential walking, two-legged hopping, diagonal sequential walking, pacing and diagonal trotting. The parameter setting method comprises the following steps: S1. Linearize the time-delay coupled nonlinear system near the equilibrium point to obtain the corresponding linear system and the corresponding characteristic equation; S2. Assume that the characteristic equation has a pair of pure imaginary roots with zero real part, substitute the assumed pure imaginary roots into the characteristic equation, separate the real part and the imaginary part, and obtain two equations of the real part and the imaginary part. Then, according to the relationship between trigonometric functions, obtain the algebraic equation corresponding to the periodic rhythm; S3. Calculate the algebraic equation to get the corresponding frequency value, substitute it into the real part and imaginary part equations to calculate the critical lag value; S4. Verify the parameter plane determined by the critical time lag, take any value in the region, and use numerical methods to verify that the region is a rhythmic mode or a resting mode. The rhythmic mode region and the resting mode region appear alternately, and the corresponding rhythmic mode region is distributed with quadrupedal hopping, lateral walking, bipedal hopping, and diagonal walking modes in sequence; By adjusting the network topology, the parameter region of bipedal hopping can be further improved to achieve pacing and diagonal trotting modes.

2. According to claim 1, a method for setting parameters of a CPG control system of a quadruped robot, It is characterized in that The unidirectional ring network structure is specifically a single ring network to realize four gait modes, and the four gait modes are divided into three cases: a. quadrupedal hopping, lateral walking, diagonal walking and bipedal hopping; b. quadrupedal hopping, lateral walking, diagonal walking and pacing; c. Four-legged hopping, sideways walking, diagonal walking and diagonal trotting.

3. The method for setting parameters of a CPG control system of a quadruped robot according to claim 2, It is characterized in that The unidirectional ring network structure is specifically a superposition of three single ring networks to realize all six gait modes.

4. The method for setting parameters of a CPG control system of a quadruped robot according to claim 1, It is characterized in that The oscillator units are connected by using a nonlinear time-delay coupling function or a linear time-delay coupling function.

5. The method for setting parameters of a CPG control system of a quadruped robot according to claim 1, It is characterized in that The oscillator unit adopts an oscillator model with a periodic rhythm signal.

6. The method for setting parameters of a CPG control system of a quadruped robot according to claim 1, It is characterized in that The oscillator unit adopts a node model without a periodic rhythm signal.

7. The method for setting parameters of a CPG control system of a quadruped robot according to claim 1, It is characterized in that The models and coupling functions adopted by the four oscillator units are the same, so that the CPG control system has symmetry.

8. The method for setting parameters of a CPG control system of a quadruped robot according to claim 1, It is characterized in that The nonlinear dynamic analysis is specifically a Hopf bifurcation analysis. Based on the Hopf bifurcation theory, the CPG control system can generate periodic rhythmic stability switching under the action of time lag.

9. The method for setting parameters of a CPG control system of a quadruped robot according to claim 8, It is characterized in that The parameter region corresponding to the periodic rhythm produces a variety of stable phase differences based on the spontaneous symmetry breaking mechanism, which is the parameter region of the gait pattern.