中文摘要：

为了探索新的CPG神经网络算法,提出了一种基于自生长神经元构建CPG神经网络的方法.首先,基于万有引力定理建立神经元生长锥的运动方程;然后,通过全局搜索算法建立神经元的目标连接算法;最后,以四足机器人为对象建立其步态控制网络,通过数值仿真实验表明,该算法可以有效地建立多种步态网络.

英文摘要：

In order to explore a new algorithm for CPG neural network, an algorithm of neural network based on self- growing of neuron is pro- posed. First of all, establishing the movement equa- tions of grown cone based on the universal gravita- tion. And then in order to establish a connection al- gorithm of the neuron the global searching algo- rithm is uesed. Finally, the paper uses the quadru- ped as the simulation object to construct the net- work of gait control. And the simulation experi- ment shows the proposed approach can creat the neural networks on a variety of gaits effectively.