14
Biologically Inspired Robotics
2.1 Introduction
With elongated and limbless bodies, as well as scales, snakes perform many
kinds of nimble motions, adapting to different kinds of environments. By
using such advantageous characteristics of snakes, snake-like robots are
expected to be applied to perform search or rescue tasks in an unstructured
environment, where traditional mobile mechanisms cannot move well.
However, it is difficult to control a snake-like robot effectively due to the
fact that the robot has high degrees of freedom. Some eel-like or snake-like
robots present elegant mechanical designs or life-like movements (see Ma
2001; McIsaac and Ostrowski 2000; Mori and Hirose 2002). These snake-like
robots were designed to imitate the shape of snakes. Most are complicated
models with numerous calculations or little environmental adaptability.
Furthermore, it is difficult to change gait patterns using simple commands.
Recently, researchers have turned their attention to bio-inspired locomotion control methods. Animals can spontaneously carry out walking, respiration, and flying without any careful planning. Most of these rhythmic
motions are controlled by a rhythm-generating mechanism, which is called
the central pattern generator (CPG) (Mattia, Paolo, and Luigi 2004). The CPG
can generate self-induced oscillation even without a high level command.
Due to the advantage of the CPG in rhythmic motion control, many researchers have placed CPG schemes into the control of robots. By implementing
CPG oscillators in the mechanism of a robot arm, more adaptive movement
was performed (Williamson 1998). A dynamic gait was achieved through
the use of a neural system model in a quadruped robot (Fukuoka, Kimura,
and Cohen 2003). Based on biomimetic CPGs and on information from distributed distance sensors, neuromuscular motion control for an undulatory
robot model was presented (Sfakiotakis and Tsakiris 2008). An amphibious
robotic snake realized a crawl motion by utilizing an onboard CPG (Crespi
and Ijspeert 2008). The generation of rhythmic and voluntary patterns of
mastication has been tested on a humanoid chewing robot by using a CPG
oscillator (Xu et al. 2009).
The rhythmic creeping motion of a snake can also be generated with a
CPG mechanism. Much research has been conducted on snake-like robotic
locomotion controlled by a CPG (see Conradt and Varshavskaya 2003; Inoue,
Ma, and Jin 2004; Matsuo, Yokoyama, and Ishii 2007). Most of these studies employed a CPG network with a unilateral open-loop connection. The
output of the CPG oscillator in this network requires an additional calculation to develop more suitable rhythmic signals for robot control due to its
irregular output wave. In order to achieve a better signal without any additional calculation, a network with feedback connection was proposed to generate rhythmic output with uniform amplitude and phase difference (Wu
and Ma 2010). The influence of each CPG parameter on the output of CPG
was also analyzed. Based on results of the analysis, the locomotion control
Biologically Inspired Robotics
2.1 Introduction
With elongated and limbless bodies, as well as scales, snakes perform many
kinds of nimble motions, adapting to different kinds of environments. By
using such advantageous characteristics of snakes, snake-like robots are
expected to be applied to perform search or rescue tasks in an unstructured
environment, where traditional mobile mechanisms cannot move well.
However, it is difficult to control a snake-like robot effectively due to the
fact that the robot has high degrees of freedom. Some eel-like or snake-like
robots present elegant mechanical designs or life-like movements (see Ma
2001; McIsaac and Ostrowski 2000; Mori and Hirose 2002). These snake-like
robots were designed to imitate the shape of snakes. Most are complicated
models with numerous calculations or little environmental adaptability.
Furthermore, it is difficult to change gait patterns using simple commands.
Recently, researchers have turned their attention to bio-inspired locomotion control methods. Animals can spontaneously carry out walking, respiration, and flying without any careful planning. Most of these rhythmic
motions are controlled by a rhythm-generating mechanism, which is called
the central pattern generator (CPG) (Mattia, Paolo, and Luigi 2004). The CPG
can generate self-induced oscillation even without a high level command.
Due to the advantage of the CPG in rhythmic motion control, many researchers have placed CPG schemes into the control of robots. By implementing
CPG oscillators in the mechanism of a robot arm, more adaptive movement
was performed (Williamson 1998). A dynamic gait was achieved through
the use of a neural system model in a quadruped robot (Fukuoka, Kimura,
and Cohen 2003). Based on biomimetic CPGs and on information from distributed distance sensors, neuromuscular motion control for an undulatory
robot model was presented (Sfakiotakis and Tsakiris 2008). An amphibious
robotic snake realized a crawl motion by utilizing an onboard CPG (Crespi
and Ijspeert 2008). The generation of rhythmic and voluntary patterns of
mastication has been tested on a humanoid chewing robot by using a CPG
oscillator (Xu et al. 2009).
The rhythmic creeping motion of a snake can also be generated with a
CPG mechanism. Much research has been conducted on snake-like robotic
locomotion controlled by a CPG (see Conradt and Varshavskaya 2003; Inoue,
Ma, and Jin 2004; Matsuo, Yokoyama, and Ishii 2007). Most of these studies employed a CPG network with a unilateral open-loop connection. The
output of the CPG oscillator in this network requires an additional calculation to develop more suitable rhythmic signals for robot control due to its
irregular output wave. In order to achieve a better signal without any additional calculation, a network with feedback connection was proposed to generate rhythmic output with uniform amplitude and phase difference (Wu
and Ma 2010). The influence of each CPG parameter on the output of CPG
was also analyzed. Based on results of the analysis, the locomotion control