CN103203746B

CN103203746B - Biped robot CPG net control topological structure construction method

Info

Publication number: CN103203746B
Application number: CN201210378285.3A
Authority: CN
Inventors: 陈启军; 刘成菊
Original assignee: Tongji University
Current assignee: Tongji University
Priority date: 2012-09-29
Filing date: 2012-09-29
Publication date: 2015-10-28
Anticipated expiration: 2032-09-29
Also published as: CN103203746A

Abstract

The present invention relates to a kind of biped robot CPG net control topological structure construction method, comprise the following steps: CPG net control is divided into the health net control (body for controlling hip joint? network) part and the leg net control (leg for controlling leg joint? network) part, the symmetry of left and right leg joint control signal and the rotational phase relation of left and right leg control in the process of walking to achieve biped robot; The mode of being of coupled connections in CPG net control between neuron elements is optimized, reduces the complexity of CPG net control; Parameter tuning is carried out to CPG net control, constructs optimal network topological structure.Compared with prior art, the present invention has the advantages such as complexity is low, net control is rational in infrastructure.

Description

Construction method of biped robot CPG control network topology

技术领域technical field

本发明涉及一种网络拓扑结构的信息处理方法，尤其是涉及一种两足机器人CPG控制网络拓扑结构构建方法。The invention relates to an information processing method of a network topology structure, in particular to a method for constructing a biped robot CPG control network topology structure.

背景技术Background technique

行走控制是两足和人形机器人研究和应用领域中的一项关键技术。传统的方法是采用基于模型的人工规划，使机器人按预先设定的运动轨迹进行运动。随着机器人逐渐应用于非结构化环境，基于机器人模型和行走环境建模的传统方法阻碍了机器人的实际应用。随着人们对两足动物步行本质的深入研究和神经科学的不断发展，基于神经科学的控制方法逐渐被应用到两足机器人的行走控制中。基于中枢模式发生器（CPG，central pattern generator）的控制思路是这一方向的一个典型代表。Walking control is a key technology in the research and application fields of biped and humanoid robots. The traditional method is to use model-based manual planning to make the robot move according to the preset trajectory. As robots are gradually applied to unstructured environments, traditional methods based on robot models and walking environment modeling hinder the practical application of robots. With the in-depth research on the nature of biped walking and the continuous development of neuroscience, control methods based on neuroscience are gradually being applied to the walking control of biped robots. The control idea based on central pattern generator (CPG, central pattern generator) is a typical representative of this direction.

生物学家认为，动物的运动控制神经网络以CPG为中心，接受来自高层神经中枢的调控命令，以及来自躯体感受器的反馈信息。CPG是由神经元构成的局部振荡网络，能够通过神经元之间的相互抑制产生稳定的相位互锁关系，并通过自激振荡激发躯体相关部位产生节律运动。大脑的高层调控和环境反馈可以对动物的节律运动起到调节作用，使动物的运动具有适应性。基于CPG的控制方法具有生物上的可解释性，因而最近在工程界引起了广泛的兴趣，并开始将CPG机理进行工程建模，应用于各类机器人的运动控制中。力图通过将生物运动神经系统的控制机理与机器人的仿生运动机构相结合，提高机器人的运动性能，促进机器人在各种实际环境中的实用化进程。Biologists believe that the animal's motor control neural network is centered on CPG, receiving regulatory commands from the high-level nerve center, and feedback information from somatoceptors. CPG is a local oscillation network composed of neurons, which can generate a stable phase interlocking relationship through mutual inhibition between neurons, and stimulate the rhythmic movement of relevant parts of the body through self-excited oscillations. The high-level regulation of the brain and environmental feedback can regulate the rhythmic movement of animals, making the animal's movement adaptive. The control method based on CPG is biologically explainable, so it has recently aroused widespread interest in the engineering community, and began to carry out engineering modeling of the CPG mechanism and apply it to the motion control of various robots. It is trying to improve the motion performance of the robot and promote the practical application of the robot in various practical environments by combining the control mechanism of the biological motor nervous system with the bionic motion mechanism of the robot.

基于CPG控制的基本思路是：首先对CPG进行工程建模，设计一个可以产生稳定振荡输出信号的函数，作为机器人自由度的控制器，多个自由度的控制一般用多个CPG单元构成的CPG网络来实现，改变网络的拓扑结构可以改变振荡信号的输出模式，从而实现不同的运动模式。目前，对于两足机器人这种复杂的系统，CPG模型并不具备很强的实用性，目前的研究也多停留在模拟仿真阶段，或者仅仅是对两足机器人某些关节的节律运动进行控制。CPG机理在机器人运动控制中的应用主要是采用关节空间（joint space）控制方法，其整体的控制构架如图1所示。在控制系统中，CPG网络模块是产生关节控制信号的核心模块，其网络拓扑结构设计的合理性和有效性关系到控制效果的优劣。目前在研究和工程应用时，一般将CPG单元按照机器人的自由度分布，逐一分配到机器人的关节空间，利用多个CPG单元之间的相互耦合产生期望的运动模式。将该控制方法应用于两足机器人目前还没有成功的实体实验，难点就在于CPG网络拓扑结构设计还没有统一的设计方法，网络结构的复杂性增加了参数整定的难度，很难产生期望的关节控制信号。另一方面，由于两足机器人自由度较多，构成控制器所需要的CPG单元数目也相对较多。如何确定CPG单元之间合理的拓扑连接及模型参数，是一个复杂的寻优问题。一般需要采用进化算法对控制系统进行优化，但由于同时涉及拓扑结构寻优和模型参数优化，采用传统的优化算法难以获得好的结果。同时，这种方法优化耗时较长，在机器人实体上实现也有一定困难。因此不能单纯的将CPG单元简单的分配给机器人的各个自由度，依赖进化计算进行寻优，必须设计合理的网络拓扑结构构架为前提，再辅以进化计算来进一步优化CPG的连接方式和模型参数。The basic idea of CPG-based control is: first, carry out engineering modeling on CPG, design a function that can generate a stable oscillating output signal, and use it as a controller for robot degrees of freedom. The control of multiple degrees of freedom generally uses a CPG composed of multiple CPG units. Network, changing the topology of the network can change the output mode of the oscillation signal, so as to achieve different motion modes. At present, for such a complex system as a biped robot, the CPG model does not have strong practicability, and most of the current research stays in the simulation stage, or only controls the rhythmic motion of some joints of the biped robot. The application of CPG mechanism in robot motion control mainly adopts joint space (joint space) control method, and its overall control framework is shown in Figure 1. In the control system, the CPG network module is the core module that generates joint control signals, and the rationality and effectiveness of its network topology design are related to the quality of the control effect. At present, in research and engineering applications, the CPG units are generally allocated to the joint space of the robot one by one according to the distribution of the degrees of freedom of the robot, and the mutual coupling between multiple CPG units is used to generate the desired motion mode. There is no successful physical experiment to apply this control method to biped robots. The difficulty lies in the fact that there is no unified design method for CPG network topology design. The complexity of the network structure increases the difficulty of parameter tuning, and it is difficult to generate the desired joints. control signal. On the other hand, due to the many degrees of freedom of the biped robot, the number of CPG units required to form the controller is also relatively large. How to determine the reasonable topological connections and model parameters between CPG units is a complicated optimization problem. Generally, an evolutionary algorithm is needed to optimize the control system, but because it involves both topology optimization and model parameter optimization, it is difficult to obtain good results by using traditional optimization algorithms. At the same time, the optimization of this method takes a long time, and it is difficult to realize it on the robot body. Therefore, it is not possible to simply assign the CPG unit to each degree of freedom of the robot and rely on evolutionary calculation for optimization. It is necessary to design a reasonable network topology structure as a premise, and then use evolutionary calculation to further optimize the connection mode and model parameters of the CPG. .

发明内容Contents of the invention

本发明的目的就是为了克服上述现有技术存在的缺陷而提供一种复杂度低、控制网络结构合理的两足机器人CPG控制网络拓扑结构构建方法。The object of the present invention is to provide a biped robot CPG control network topology construction method with low complexity and reasonable control network structure in order to overcome the above-mentioned defects in the prior art.

本发明的目的可以通过以下技术方案来实现：The purpose of the present invention can be achieved through the following technical solutions:

一种两足机器人CPG控制网络拓扑结构构建方法，包括以下步骤：A method for constructing a biped robot CPG control network topology, comprising the following steps:

1）将CPG控制网络分为用于控制机器人髋关节的身体控制网络（bodynetwork）部分和用于控制腿部关节的腿部控制网络（leg network）部分，便于控制双足机器人左右腿控制信号的对称性和机器人在行走过程中左右腿的相位关系；1) The CPG control network is divided into the body control network (body network) part for controlling the robot hip joint and the leg control network (leg network) part for controlling the leg joints, which is convenient for controlling the left and right leg control signals of the biped robot Symmetry and the phase relationship between the left and right legs of the robot during walking;

2）对CPG网络内神经元单元之间的耦合方式进行优化，简化了神经元之间的连接方式；2) Optimize the coupling mode between neuron units in the CPG network, simplifying the connection mode between neurons;

3）对CPG控制网络的进行参数整定，构建出最优网络拓扑结构。3) Adjust the parameters of the CPG control network to build an optimal network topology.

步骤1）中的CPG控制网络采用神经元振荡器模型，该模型的数学表达式为：The CPG control network in step 1) adopts the neuron oscillator model, and the mathematical expression of the model is:

${T T}_{r r} {\overset{\cdot &Center Dot;}{u u}}_{i i}^{{{e e,, f f}}} = = {- - u u}_{i i}^{{{e e,, f f}}} + + {w w}_{fe fe} {r r}_{i i}^{{{f f,, e e}}} - - β β {v v}_{i i}^{{{e e,, f f}}} + + {s the s}_{00} + + {Feed Feed}_{i i}^{{{e e,, f f}}} + + {Σ Σ}_{j j = = 11}^{n no} {w w}_{ij ij} {r r}_{j j}^{{{e e,, f f}}}$

${T T}_{a a} {\overset{\cdot &Center Dot;}{v v}}_{i i}^{{{e e,, f f}}} = = - - {v v}_{i i}^{{{e e,, f f}}} + + {r r}_{i i}^{{{e e,, f f}}}$

${r r}_{i i}^{{{e e,, f f}}} = = max max (({u u}_{i i}^{{{e e,, f f}}},, 00))$

${r r}_{i i} = = {- - u u}_{i i}^{{{e e}}} + + {u u}_{i i}^{{{f f}}}$

其中，i表示第i个CPG单元，e表示屈肌神经元，f表示伸肌神经元，u_i为神经元的内部状态，v_i为神经元自抑制状态，为神经元的输出，T_r和T_a分别为上升时间和适应时间常数，w_fe为神经元的相互抑制系数，β为神经元的自抑制系数，s₀代表运动控制网络输出的周期性振荡信号，Feed_i为反馈输入信号，w_ij为神经元j与神经元i间的连接权重，r_i为第i个CPG单元的输出，为了消除振荡器输出信号为零部分的影响，本设计采用屈肌神经元、伸肌神经元的状态项线性合成振荡器的输出（Kimura模型利用神经元的输出项和线性合成振荡器的输出）。Among them, i represents the i-th CPG unit, e represents the flexor neuron, f represents the extensor neuron, u _i is the internal state of the neuron, v _i is the neuron self-inhibition state, is the output of the neuron, T _r and T _a are the rise time and the adaptation time constant respectively, w _fe is the mutual inhibition coefficient of the neuron, β is the self-inhibition coefficient of the neuron, s ₀ represents the periodic oscillation of the motor control network output signal, Feed _i is the feedback input signal, w _ij is the connection weight between neuron j and neuron i, ri is the output of the _i -th CPG unit, in order to eliminate the influence of the oscillator output signal being zero, this design uses State items for flexor neurons and extensor neurons The output of the linear synthesis oscillator (the Kimura model uses the output term of the neuron and output of the linear synthesis oscillator).

步骤1）中在构建CPG网络拓扑结构时，将两足机器人的髋关节和腿部的其他关节分别设计身体控制网络（body network）和腿部控制网络（leg network）来控制，目的是在保证两足机器人左右腿控制信号的对称性和机器人在行走过程中左右腿的相位关系的同时，减低系统的复杂度，减少需要优化的网络参数。In step 1), when constructing the CPG network topology, design the body control network (body network) and the leg control network (leg network) to control the hip joints and other joints of the legs of the biped robot respectively, in order to ensure The symmetry of the control signals of the left and right legs of the biped robot and the phase relationship between the left and right legs of the robot during walking can reduce the complexity of the system and reduce the network parameters that need to be optimized.

步骤2）中具体的优化方法为：仅考虑屈肌神经元之间以及伸肌神经元之间的相互抑制系数来获取关节自由度之间的相位关系。The specific optimization method in step 2) is: only consider the mutual inhibition coefficient between flexor neurons and extensor neurons to obtain the phase relationship between joint degrees of freedom.

步骤3）具体包括以下步骤：Step 3) specifically includes the following steps:

A）采用单参数分析法获取CPG控制网络的单个模型参数对于关节控制信号的影响趋势；A) Using the single parameter analysis method to obtain the influence trend of a single model parameter of the CPG control network on the joint control signal;

B）根据每个模型参数对关节控制信号的影响趋势并辅助多目标进化计算方法，获取可以使机器人在平地行走的最优关节控制信号；B) According to the influence trend of each model parameter on the joint control signal and assisting the multi-objective evolutionary calculation method, the optimal joint control signal that can make the robot walk on the flat ground is obtained;

C）根据实际的行走效果，对步骤B）中最优关节控制信号下的模型参数进行微调。C) According to the actual walking effect, fine-tune the model parameters under the optimal joint control signal in step B).

与现有技术相比，本发明基于CPG的两足机器人的空间控制方法，提出了一种新的CPG控制网络构建方法，并设计了更加合理有效的模型参数整定策略。本设计在一定程度上改进了传统的控制网络构建方法，使得其复杂度降低，控制网络结构更加合理。Compared with the prior art, the present invention proposes a new CPG control network construction method based on the CPG-based biped robot space control method, and designs a more reasonable and effective model parameter tuning strategy. This design improves the traditional control network construction method to a certain extent, which reduces its complexity and makes the control network structure more reasonable.

附图说明Description of drawings

图1为关节空间控制方法的控制系统整体构架；Fig. 1 is the overall structure of the control system of the joint space control method;

图2为两足机器人的关节自由度分布图；Figure 2 is a distribution diagram of the joint degrees of freedom of the biped robot;

图3为用于控制髋关节的身体控制网络（body network）部分的CPG网络结构；Figure 3 is the CPG network structure of the body network (body network) part used to control the hip joint;

图4为身体控制网络（body network）部分输出的振荡信号；Figure 4 is the oscillating signal output by the body network part;

图5为由本发明提出的CPG控制网络拓扑结构；Fig. 5 is the CPG control network topology structure proposed by the present invention;

图6为屈肌神经元和伸肌神经元之间的耦合连接方式。Figure 6 shows the coupling connection between flexor neurons and extensor neurons.

具体实施方式Detailed ways

下面结合附图和具体实施例对本发明进行详细说明。The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

实施例Example

步骤1：图2所示是目前最常见的研究和应用的两足人形机器人腿部的自由度分布示意图。对于该类型的两足机器人行走控制，髋关节自由度对左右两腿相位的控制最为重要。因此本设计将CPG控制网络分为用于控制髋关节的身体控制网络（body network）部分和用于控制腿部关节的腿部控制网络（leg network）部分，实现双足机器人左右腿控制信号的对称性和机器人在行走过程中左右腿的相位关系。以身体控制网络（body network）部分为例，如图3所示，通过四个神经元单元的双向耦合连接构成身体控制网络（body network）来控制两足机器人髋关节的四个自由度。将身体控制网络（body network）部分的四个神经元单元之间采用等权重的相互抑制连接（抑制权重设为-1），可以得到如图4所示的相位关系，左右腿的HipPitch相位差为180°，满足左右髋关节前后方向自由度控制信号的反相关系，同时能得到同一个关节的两个自由度的理想的90°的相位差关系。Step 1: Fig. 2 is a schematic diagram of the degree of freedom distribution of the legs of the most commonly researched and applied biped humanoid robot. For the walking control of this type of biped robot, the degree of freedom of the hip joint is most important for the phase control of the left and right legs. Therefore, in this design, the CPG control network is divided into the body control network (body network) part for controlling the hip joint and the leg control network (leg network) part for controlling the leg joints, so as to realize the control signal of the left and right legs of the biped robot. Symmetry and phase relationship of the left and right legs of the robot during walking. Taking the body network part as an example, as shown in Figure 3, the body network is composed of bidirectional coupling connections of four neuron units to control the four degrees of freedom of the hip joint of the biped robot. The four neuron units in the body network part are connected by mutual inhibition with equal weights (the inhibition weight is set to -1), and the phase relationship shown in Figure 4 can be obtained. The HipPitch phase difference between the left and right legs It is 180°, which satisfies the anti-phase relationship of the control signals of the front and rear direction freedom of the left and right hip joints, and at the same time can obtain the ideal 90° phase difference relationship of the two degrees of freedom of the same joint.

此处，CPG控制网络采用神经元振荡器模型，该模型的数学表达式为：Here, the CPG control network adopts the neuron oscillator model, and the mathematical expression of the model is:

${T T}_{r r} {\overset{\cdot \cdot}{u u}}_{i i}^{{{e e,, f f}}} = = {- - u u}_{i i}^{{{e e,, f f}}} + + {w w}_{fe fe} {r r}_{i i}^{{{f f,, e e}}} - - β β {v v}_{i i}^{{{e e,, f f}}} + + {s the s}_{00} + + {Feed Feed}_{i i}^{{{e e,, f f}}} + + {Σ Σ}_{j j = = 11}^{n no} {w w}_{ij ij} {r r}_{j j}^{{{e e,, f f}}}$

${r r}_{i i}^{{{e e,, f f}}} = = max max (({u u}_{i i}^{{{e e,, f f}}},, 00))$

${r r}_{i i} = = {- - u u}_{i i}^{{{e e}}} + + {u u}_{i i}^{{{f f}}}$

步骤2：对CPG控制网络内神经元之间的耦合连接方式进行优化，降低CPG网络的复杂度。传统的神经元之间的连接方式是屈肌神经元和伸肌神经元全向耦合，对于两个神经元单元，需要考虑四个连接权重的调制。对于多个自由度的两足机器人，如果按照传统的连接方式，需要优化的连接参数数量庞大，很难获取合适的优化结果。本发明中采用如图6所示的耦合连接方法：考虑屈肌神经元之间以及伸肌神经元之间的相互抑制系数便可以得到期望的相位关系，因此为了降低整个拓扑结构的复杂度。Step 2: Optimize the coupling connection mode between neurons in the CPG control network to reduce the complexity of the CPG network. The traditional connection between neurons is the omnidirectional coupling of flexor neurons and extensor neurons. For two neuron units, the modulation of four connection weights needs to be considered. For biped robots with multiple degrees of freedom, if the traditional connection method is used, the number of connection parameters that need to be optimized is huge, and it is difficult to obtain suitable optimization results. In the present invention, the coupling connection method shown in FIG. 6 is adopted: the desired phase relationship can be obtained by considering the mutual inhibition coefficients between flexor neurons and extensor neurons, so in order to reduce the complexity of the entire topology.

步骤3：基于进化计算对CPG控制网络的进行参数整定，优化CPG控制网络的模型参数和拓扑连接方式，具体可细分为以下几步：Step 3: Based on the evolutionary calculation, the parameters of the CPG control network are adjusted, and the model parameters and topological connection mode of the CPG control network are optimized, which can be subdivided into the following steps:

首先，采用单参数分析法获取CPG控制网络的单个模型参数（T_r，T_a，β，s₀和w_fe）对于关节控制信号的影响趋势。由于参数之间是相互耦合，参数对输出的影响不存在线性关系，选定其中一个模型参数作为被测参数，仅改变该被测参数的值，通过判断CPG输出信号是否稳定振荡，预估被测参数的取值范围；在预估的范围内连续调节被测参数，测定输出的可量化表征量（幅值、频率、相位等）与被测参数的定量关系；First, the influence trend of single model parameters (T _r , T _a , β, s ₀ and w _fe ) of the CPG control network on the joint control signal was obtained by single-parameter analysis. Since the parameters are coupled with each other, there is no linear relationship between the parameters and the output. Select one of the model parameters as the measured parameter, and only change the value of the measured parameter. By judging whether the CPG output signal is stable and oscillating, it is estimated The value range of the measured parameter; continuously adjust the measured parameter within the estimated range, and determine the quantitative relationship between the quantifiable characterization of the output (amplitude, frequency, phase, etc.) and the measured parameter;

然后，基于进化计算方法获得能使机器人平地行走的关节控制信号。基于CPG的关节空间控制方法由于不基于两足机器人的运动学模型和逆运动学计算，所以机器人行走的稳定性只能通过优化CPG网络的参数，得到能实现机器人各个自由度协调动作的关节控制信号。为了提高机器人的行走稳定性，本设计在参数优化目标中考虑利用机器人行走时的稳定裕度（将ZMP到支撑凸多边形边界的最短距离作为步行系统的稳定裕度）来提高机器人行走的稳定性。本设计利用多目标进化算法，基于Kalyanmoy Deb的NSGA-II（Non-Dominated Sorting Genetic Algorithm-II）进行多目标遗传算法的设计。优化的目标是得到能实现机器人平地行走的最优拓扑连接方式和CPG的模型参数，适应度函数的设计考虑机器人行走的可行性和稳定性，设计为：Then, the joint control signals that can make the robot walk on flat ground are obtained based on the evolutionary calculation method. Since the joint space control method based on CPG is not based on the kinematics model and inverse kinematics calculation of the biped robot, the stability of the robot's walking can only be obtained by optimizing the parameters of the CPG network to obtain the joint control that can realize the coordinated movements of the robot's various degrees of freedom. Signal. In order to improve the walking stability of the robot, this design considers the stability margin when the robot is walking (the shortest distance from the ZMP to the boundary of the supporting convex polygon is used as the stability margin of the walking system) in the parameter optimization goal to improve the walking stability of the robot . This design uses multi-objective evolutionary algorithm, based on Kalyanmoy Deb's NSGA-II (Non-Dominated Sorting Genetic Algorithm-II) for the design of multi-objective genetic algorithm. The goal of optimization is to obtain the optimal topological connection mode and CPG model parameters that can realize robot walking on flat ground. The design of fitness function considers the feasibility and stability of robot walking, and is designed as:

${fitness fitness}_{11} = = 11 / / \sqrt{{(({x x}_{end end} - - {x x}_{00}))}^{22}}$

fitness₂＝1/D_s fitness ₂ = 1/D _s

其中，x₀是机器人的初始位置（x方向），x_end代表第机器人停止行走时的位置（x方向）。第一个目标函数是为了实现机器人的直线行走，因此在优化时只考虑机器人前进（x方向）方向行走距离。第二个优化目标考虑行走的稳定性，D_s为机器人行走的稳定裕度。Among them, x ₀ is the initial position of the robot (x direction), and x _end represents the position (x direction) of the robot when it stops walking. The first objective function is to realize the straight-line walking of the robot, so only the walking distance of the robot in the forward direction (x direction) is considered during optimization. The second optimization objective considers the stability of walking, and D _s is the stability margin of robot walking.

最后，根据两足机器人实际的行走效果，对步骤B）中最优关节控制信号下的模型参数进行微调，对其进行优化。最后，根据实际的机器人行走实验效果，在完成步骤A）和步骤B）的基础上结合设计者的经验来综合调整参数。该参数整定方法充分结合了计算智能和人类的智能，让设计者可以根据优化算法的综合输出结果，逐步加深对控制器和被控对象的了解，并获得新的设计思想和灵感。Finally, according to the actual walking effect of the biped robot, the model parameters under the optimal joint control signal in step B) are fine-tuned and optimized. Finally, according to the actual robot walking experiment effect, the parameters are adjusted comprehensively on the basis of completing step A) and step B) combined with the experience of the designer. This parameter tuning method fully combines computational intelligence and human intelligence, allowing designers to gradually deepen their understanding of the controller and the controlled object based on the comprehensive output results of the optimization algorithm, and obtain new design ideas and inspirations.

本发明基于CPG的两足机器人的空间控制方法，提出了一种新的CPG控制网络构建方法，并设计了更加合理有效的模型参数整定策略，在一定程度上改进了传统的控制网络构建方法，使得其复杂度降低，控制网络结构更加合理。The present invention proposes a new CPG control network construction method based on the space control method of the CPG biped robot, and designs a more reasonable and effective model parameter setting strategy, which improves the traditional control network construction method to a certain extent. It reduces its complexity and makes the control network structure more reasonable.

Claims

1. a biped robot CPG net control topological structure construction method, is characterized in that, comprise the following steps:

1) CPG net control be divided into the part of the body network for controlling hip joint and be used for controlling the leg network part in leg joint, ensureing the phase relation of the symmetry of biped robot left and right leg control signal and robot left and right leg in the process of walking;

2) mode of being of coupled connections between CPG network introflexion muscular nerve unit and extensor neuron is optimized, reduces the complexity of CPG net control;

3) parameter tuning is carried out to CPG net control, constructs optimal network topological structure, specifically comprise the following steps:

A) single-parameter analysis method is adopted to obtain the effect tendency of single model parameter for joint control signal of CPG net control;

B) according to each model parameter, also multi-target evolution computational methods are assisted to the effect tendency of joint control signal, obtain and robot can be made in the optimal joint control signal of level walking;

C) according to the walking effect of reality, to step B) in model parameter under optimal joint control signal finely tune;

Step 2) in concrete optimization method be: only consider that mutual rejection coefficient between musculus flexor neuron and between extensor neuron is to obtain the phase relation between joint freedom degrees.

2. a kind of biped robot CPG net control topological structure construction method according to claim 1, is characterized in that, step 1) in CPG network adopt neural oscillator model, the mathematic(al) representation of this model is:

T_{r} {\overset{\cdot}{u}}_{i}^{{e, f}} = - u_{i}^{{e, f}} + w_{f e} r_{i}^{{f, e}} - {βv}_{i}^{{e, f}} + s_{0} + {Feed}_{i}^{{e, f}} + Σ_{j = 1}^{n} w_{i j} r_{j}^{{e, f}}

T_{a} {\overset{\cdot}{v}}_{i}^{{e, f}} = - v_{i}^{{e, f}} + r_{i}^{{e, f}}

r_{i}^{{e, f}} = m a x (u_{i}^{{e, f}}, 0)

r_{i} = - u_{i}^{{e}} + u_{i}^{{f}}

Wherein, i represents i-th CPG unit, and e represents musculus flexor neuron, and f represents extensor neuron, u _ifor neuronic internal state, v _ifor neuron is from holddown, r _i ^{{ e, f}}for neuronic output, T _rand T _abe respectively rise time and adaptation time constant, w _fefor neuronic mutual rejection coefficient, β is neuronic from rejection coefficient, s ₀represent the periodic swinging signal that Motion Control Network exports, Feed _ifor feedback input signal, w _ijfor the connection weight between neuron j and neuron i, r _ibe the output of i-th CPG unit, in order to oscillation-damped device output signal is the impact of null part, adopt musculus flexor neuron, the neuronic status items of extensor the output of linear synthesis oscillator.

3. a kind of biped robot CPG net control topological structure construction method according to claim 1, it is characterized in that: the hip joint of biped robot and leg joint are designed respectively body network part and legnetwork partly controls, while the phase relation ensureing the symmetry of biped robot left and right leg control signal and robot left and right leg in the process of walking, the complexity of attenuating system, reduces the network parameter needing to optimize.