Biomedical Engineering Reference
In-Depth Information
Figure 16.13
Simulation of a robot's force control while interacting with its unknown dynamic environment.
15
1
10
5
0
0
m
e
+m
r
d
e
+d
r
k
e
+k
r
d
e
k
e
−
5
1
−
−
10
−
2
−
15
0
5
10
15
Time [s]
20
25
30
0
5 0
15
Time [s]
0 5 0
(b)
Figure 16.14 (See color insert following page 302) Simulation results on the time responses of the robot's
force (the blue lines) tracking a time varying desired force (the green lines), and the parameter convergences of the
feedforward controller. The environmental viscosity is set from 2 to 0.5 at the time t = 250[s] in (a) and t = 15[s] in (b).
At the beginning 100[s], 250[s] to 350[s] in [a] and the beginning 4[s] in (b), the desired forces are set as noises.
the same reaction force. It is also clear that when the environment changes its dynamics (at
simulation time
t
= 250[
s
] in Figure 16.14(a) or
t
= 15[
s
] in (b)), only if the desired forces satisfy
the PE condition, we can obtain the new inverse of the force controlled object. Therefore, the
adaptation (16.40) or (16.41) has the so-called generality as we discussed in the learning theory of
artificial neural networks.
Search WWH ::
Custom Search