The Application Of Central Pattern Generator In The Locomotion Control Of A Hexapod Robot

Nowadays,robot technology has been developing at a really fast speed.As a typical member of the robot family,the hexapod robot,which acts as a six-legged robot,has been applied in many fields,like disaster relief,surface detection,space exploration and so on.As a significant search field in robot technology,locomotion control has drawn much attention those years in control science world and engineering field.With the developing research in biological technology field,bionic control method has shown much priority in imitating the regular locomotion behavior of real animals comparing with other traditional control methods.Among all these bionic control methods,the Central pattern generator control method is the most typical one,this method could generate a regular and stable output wave without receiving higher control signals and exterior feedbacks.The CPG control method does not need to build an exact model of the environment,the algorithm is simple and efficient and could imitate gate behavior of real animals,so in this thesis,we would focus on the CPG control method and take the hexapod robot as a plant to give a detailed discussion of the robot locomotion control method.At the very beginning,some fundamental concepts were given,such as the gait period,gate length,and the ratio of stance phase.We take a single leg as an example to analysis geometric relation and give the positive and inverse kinematics calculation method.On the basis of this,we give polynomial fitting method to both stance phase and swing phase,and complete the kinematics planning to ensure the accuracy of the locomotion control.Meanwhile we choose the DC motor to drive the joint to move,and accomplish the process of mathematical modeling.After this,we build a three closed loop control system to ensure the accuracy of the trajectory tracking of the high-level output signals.Next,the thesis gives the method to generate a series of signals of the robot's angle curve by introducing the method based on a single nonlinear oscillator instead of the traditional method built on modeling.We choose the Hopf oscillator model as the basic structure of the CPG,and analysis the impact of the changing parameters of the model.Giving a series of fixed-difference-phase signals provided by the low-pass filters,and identify the switching point of the stance phase and swing phase by setting the threshold line.Ensuring that the joint control signals and the original phase signal has been related by the way of adjusting the sampling frequency and compensating for the phase difference.In this way,we could describe the integrate process of joint signal generation method.After that,aiming at the problems caused by the single-oscillator gait generation method,the thesis puts forward corresponding methods to give improvement.By selecting the modified Hopf oscillator,the frequency of swing phase and stance phase can be set separately.The accumulation of errors caused by the low-pass filters is eliminated by using CPG ring network structure.The process of locomotion control is simplified and the performance of bionic control method has also been improved successfully.Finally,we take some typical gaits as an example to make sure that the algorithm we presented could meet the correctness and Applicability standards on the six-legged robot experiment platform.