Earthworm-like robots have excellent locomotion capability in confined environments. Central Pattern Generator (CPG) based controllers utilize the dynamics of coupled nonlinear oscillators to spontaneously generate actuation signals for all segments, which offer significant merits over conventional locomotion control strategies. Thre are a number of oscillators that can be exploited for CPG control, while their performance in controlling peristaltic locomotion has not been systematically evaluated. To advance the state of the art, this study comprehensively evaluates the performance of four widely used nonlinear oscillators-Hopf, Van der Pol, Matsuoka, and Kuramoto-in controlling the planar locomotion of metameric earthworm-like robots. Specifically, the amplitude and phase characteristics of the continuous control signals used by the robot for achieving rectilinear, sidewinding, and arcuate locomotion are first summarized. On this basis, the sufficient parametric conditions for the four oscillator networks to generate the corresponding control signals are derived. Using a six-segment earthworm-like robot prototype as a platform, experiments confirm that the signals output by these oscillator networks can effectively control the robot to achieve the specified planar motion. Furthermore, the effects of the output signal waveforms of different oscillator networks on locomotion trajectories and performance metrics, as well as the effects of transient dynamics on the smoothness of gait transitions when the parameters are varied, are analyzed. The results demonstrate that their applicability varies in terms of locomotion efficiency, trajectory modulation, and smooth gait transitions. The Matsuoka oscillator lacks explicit rules for parameter modulation, the Van der Pol oscillator is advantageous in enhancing the average speed and turning efficiency, and the Hopf and Kuramoto oscillators are advantageous in terms of smooth gait transition. These findings provide valuable insights into the selection of appropriate oscillators in CPG-based controllers and lay the foundation for future CPG-based adaptive control of earthworm-like robots in complex environments.
Read full abstract