An empirical exploration of phase resetting for robust biped locomotion with dynamical movement primitives

Abstract

Abstract-We propose a framework for learning biped locomotion using dynamical movement primitives based on nonlinear oscillators. In our previous work, we suggested dynamical movement primitives as a central pattern generator (CPG) to learn biped locomotion from demonstration. We introduced an adaptation algorithm for the frequency of the oscillators based on phase resetting at the instance of heel strike and entrainment between the phase oscillator and mechanical system using feedback from the environment.In this paper, we empirically explore the role of phase resetting in the proposed algorithm for robust biped locomotion. We demonstrate that phase resetting contributes to robustness against external perturbations and environmental changes by numerical simulations and experiments with a physical biped robot.

Event-based walking control — From neurobiology to biped robots

“…[11]) and experiment (e.g. [12]) that this phase reset can improve the stability of a CPG-controlled walking machine.…”

Creating and modulating rhythms by controlling the physics of the body

“…Methods corresponding to the former case model entrainment by extracting the periodic and rhythmic patterns so that the specific waveforms are learned; e.g., the adaptive phase frequency technique by Ijspeert and colleagues (Ijspeert et al 2003), and also (Schaal 2003;Buchli et al 2006;Nakanishi et al 2004). Conversely, those corresponding to the latter case create entrainment by exploiting the body dynamics; e.g., the morphological computation by Pfeifer and colleagues cf.…”

Controlling swimming and crawling in a fish robot using a central pattern generator

“…Most of these robots are controlled using traditional control methods that combine (algorithmic) sine-based trajectory generators, and PID feedback controllers. Recently, the concept of CPGs is increasingly used as an alternative approach for online rhythmic trajectory generation (Wilbur et al, 2002;Fukuoka et al, 2003;Nakanishi et al, 2004;Ijspeert et al, 2005). In most cases, the CPGs are implemented as recurrent neural networks or systems of coupled nonlinear oscillators.…”