22
Biologically Inspired Robotics
joint angles. As stated in Section 2.3.1, the phase differences of CPG outputs
are homogeneously distributed in one period if the network is a closed-loop
type. From Equation (2.5), the sum of the total phase differences of a CPG’s
outputs can be derived by (n − 1)Φ where n is the numbers of CPGs. Because
one locomotive S-shape can be obtained by a group of rhythmic signals with
total 2π, the number of the locomotive S-shapes, N, can be given by
(n − 1)Φ n − 1
N =
2π
=
r
(2.7)
Therefore, the number of the locomotive S-shape can be varied by changing
the connection of rth CPG module to the first CPG module.
2.4 CPG-Controlled Snake-Like Robot
To verify the proposed CPG-based control method, a simulator for a snake-like
robot has been developed in an open dynamics engine (ODE) environment,
as shown in Figure 2.6. In the simulation, the interaction between the robot
and the ground is modeled with asymmetric friction by using a larger normal friction coefficient μ N and a smaller tangential friction coefficient μ T . To
realize this kind of friction model, a passive wheel is utilized for each link
of the snake-like robot. The actuators are installed on the joints of the robot
to make each joint swing from side to side, like the behavior of a snake. The
physical parameters of our snake-like robot platform are given in Table 2.3.
In the experiment, a closed-loop network with a feedback connection using a
cyclic inhibitory CPG model was selected as the oscillation generator for the
control of the snake-like robot. The output of the ith CPG was implemented
Joint
Passive
wheel
Body link
θ
FIGURE 2.6
Simulation platform of the snake-like robot.
Biologically Inspired Robotics
joint angles. As stated in Section 2.3.1, the phase differences of CPG outputs
are homogeneously distributed in one period if the network is a closed-loop
type. From Equation (2.5), the sum of the total phase differences of a CPG’s
outputs can be derived by (n − 1)Φ where n is the numbers of CPGs. Because
one locomotive S-shape can be obtained by a group of rhythmic signals with
total 2π, the number of the locomotive S-shapes, N, can be given by
(n − 1)Φ n − 1
N =
2π
=
r
(2.7)
Therefore, the number of the locomotive S-shape can be varied by changing
the connection of rth CPG module to the first CPG module.
2.4 CPG-Controlled Snake-Like Robot
To verify the proposed CPG-based control method, a simulator for a snake-like
robot has been developed in an open dynamics engine (ODE) environment,
as shown in Figure 2.6. In the simulation, the interaction between the robot
and the ground is modeled with asymmetric friction by using a larger normal friction coefficient μ N and a smaller tangential friction coefficient μ T . To
realize this kind of friction model, a passive wheel is utilized for each link
of the snake-like robot. The actuators are installed on the joints of the robot
to make each joint swing from side to side, like the behavior of a snake. The
physical parameters of our snake-like robot platform are given in Table 2.3.
In the experiment, a closed-loop network with a feedback connection using a
cyclic inhibitory CPG model was selected as the oscillation generator for the
control of the snake-like robot. The output of the ith CPG was implemented
Joint
Passive
wheel
Body link
θ
FIGURE 2.6
Simulation platform of the snake-like robot.