58
Other values move the system onto a

result in a number of successful steps but eventually fall.

successful walking limit cycle. Some walks don't fall despite seemingly chaotic motion.

4. 1

Up Vector Regulation Variables

The first results we show use the base PCG of Figure 3.7and the stance hip perturbations

described in Section 3.6.

The RVs consist of the components of the up vector, as illustrated in

Figure 3.12 (c).

Torso servoing is not applied.

Figure 4.2shows a representative set of RV

trajectories corresponding to

successful

walking

trials

of

60

steps.

These

curves

are

the

experimentally-based equivalents of the idealized cyclic trajectories shown earlier, such as those of

Figure 3.3.

chaotic.

In three of the plots, a clear limit cycle emerges.

In the fourth, the trajectory is

(a) F-L sampling, Q^d= [.25,0]

(b) L-F sampling, Q^d= [.2,0]

(c) SP sampling, Q^d= [.3,0]

(d) L-F sampling, Q^d= [.35,0]

Figure 4.2-

Continuous-time
diagrams

up-vector

RV

component

phase