<p>In this paper, a new piecewise linear (PWL) FitzHugh-Nagumo neural control unit for general passive actuators is presented. The PWL formulation enables unprecedented modulation of the main dynamics of the ion channels, allowing easy control of both the amplitude and period of the neuron spiking activity or the position and quality of the phase plane equilibria. Exploiting the similarities between the first-order recovery dynamics of the FitzHugh–Nagumo (FHN) model and general passive actuators, the substitution of the electrical neural recovery activity with the actuator dynamics, independently of the actuator specific nature, is presented. This realizes a novel embodied neural-motor unit. In the context of locomotion control through Central Pattern Generators, this unit introduces a number of advantages, such as reduced control order, easy state control of the actuator, modulation and adaptation capabilities. Theoretical analysis is accompanied to experimental results on the locomotion control of a robotic hand to demonstrate the effectiveness of the proposed architecture.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Locomotion control through embodied spiking oscillators

  • Alberto Motta,
  • Luca Patanè,
  • Paolo Arena

摘要

In this paper, a new piecewise linear (PWL) FitzHugh-Nagumo neural control unit for general passive actuators is presented. The PWL formulation enables unprecedented modulation of the main dynamics of the ion channels, allowing easy control of both the amplitude and period of the neuron spiking activity or the position and quality of the phase plane equilibria. Exploiting the similarities between the first-order recovery dynamics of the FitzHugh–Nagumo (FHN) model and general passive actuators, the substitution of the electrical neural recovery activity with the actuator dynamics, independently of the actuator specific nature, is presented. This realizes a novel embodied neural-motor unit. In the context of locomotion control through Central Pattern Generators, this unit introduces a number of advantages, such as reduced control order, easy state control of the actuator, modulation and adaptation capabilities. Theoretical analysis is accompanied to experimental results on the locomotion control of a robotic hand to demonstrate the effectiveness of the proposed architecture.