1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135

48

IMAGE Imgs/thesis.final.w677.gif

iii

Figure 3.19 - Estimate of desired lateral RV components
(based on S6 from Figure 3.7 (a) - rear view)
i) up vector (vertical)
ii) stance-COM
iii) swing-COM


Constructing and Applying the Discrete System Model

3. 7. 2

Once an RV target has been chosen, the discrete system model is reconstructed and applied at each

step as shown in Figure 3.20.

In general, a single fixed model is not sufficient to represent all

possible steps.

While RVs provide a reduced, low dimensional representation of system state,

they are also an incomplete and ambiguous representation of the system state.

For example, the

chosen RVs use position information and no velocity information.

Such unobserved parts of the

system state cause variations in both Qnomand J.


Figure 3.21 illustrates this idea.


This result is

evident in graphs of Figure 3.16 and Figure 3.17 where both the average slope and the offset of

the curves can vary from one step to the next.

lateral control dimension.

These variations are especially apparent in the

IMAGE Imgs/thesis.final.w678.gif

Figure 3.20 - Balancing process for each step.

[CONVERTED BY MYRMIDON]