A CPG Locomotion Control Method Of A Snake Robot Based On Oscillators

Compared with traditional legged robots,snake robots have many advantages of small cross section,uniform distribution of center of gravity and high stability.They can play a crucial role in search and rescue operations in disaster,equipment inspection,exploration,medical treatment,military task and so on.However,there still exist many problems to be solved urgently in the field of snake robots,including complex mechanical structure design,a very large number of degrees of freedom that have to be controlled and low payload capacity.The important point discussed in the paper is how to realize the locomotion of a snake robot efficiently.A locomotion control strategy based on the Central Pattern Generator(CPG)is a control method to mimic the self-excited behavior of the low-level nerve center of the biological system which is stable,easy to adjust and adaptable.In this paper,a CPG inspired control method of a snake robot based on oscillators is proposed and the main contents of the paper are as follows:1.The relationship between model parameters of the CPG networks often constructed by Matsuoka oscillators to realize the serpentine locomotion of snake robots and output waveforms is complicated and it costs much time in parametric analysis.Especially,it is hidden for the phase parameters in the model that have a significant effect on the gait of the snake robots.To address these issues,Hopf oscillators featured by limit cycle are used to construct a single-chain CPG control network,which can provide multi-dimensional yaw angle output signal for the high degree of freedom joints of a snake robot during the serpentine locomotion.The proposed CPG network has explicit frequency,amplitude and phase parameters,which can make the upper operator control directly the performance of the snake robot such as the forward speed and the curvature of the body S wave by adjusting the small amount of effective model parameters.Simulation experiments of the serpentine locomotion and the turn locomotion to avoid obstacles are performed and the experiments on a snake robot prototype verify further that the CPG model is effective in the motion control of the snake robot during the serpentine locomotion.2.In order to achieve the locomotion of the snake robot in three-dimensional space,the CPG model needs to be designed for generating three-dimensional gait signal in the complex space.Because of the structure characteristic of the snake robot prototype that each joint has two degrees of freedom,a double-chain CPG control network is constructed using Hopf oscillators as the minimum CPG unit.The convergence stability of the CPG model is proved by the contraction theory.The proposed control network can generate not only rhythmic signals of three dimensional gaits such as sidewinding locomotion and C-shape locomotion,but also produce the travelling wave signal on a plane by adjusting coefficients.Besides,a smooth gait switching method is proposed according to the movement characteristics between the adjacent joints during the serpentine locomotion and the sidewinding locomotion.The continuous adjustment of the phase difference parameters in the model can make the snake robot achieve natural gait conversion between two opposite directions and the conversion time can be arbitrarily selected.Finally,the effectiveness of the control method based on the double-chain CPG model in the snake robot locomotion control is verified by the simulation and the real experiments.3.The state parameters in the CPG model affect the locomotion patterns and characteristics of the snake robot directly.In order to improve the motion efficiency of the snake robot,a new hybrid optimization algorithm combining particle swarm optimization algorithm with rain forest algorithm is presented to find the appropriate CPG model parameters.The proposed algorithm uses adaptive partition and adjustment scales during the phase of the rainforest algorithm and introduces a relatively small population of individual particles of particle swarm by selecting appropriate stages and regions.The performance of the new algorithm is verified by four bemark functions.Finally,the simulation results show that the optimized amplitude and phase parameters are compared with the other three optimization algorithms.Finally,it shows that the algorithm can find the optimized amplitude and phase parameters by taking the simulation experiment of serpentine locomotion as an example and the proposed algorithm are further compared with other three optimization algorithms.