4

• The approach should work for creatures of reasonable complexity without making any

fundamental assumptions about the creature's structure. In particular, it should at least be

suitable for animating a human model with tens of degrees of freedom (DOFs).

• The control representation should be relatively compact, and flexible enough to allow

straightforward specification of walking variations.

• If desired, the motion should exhibit autonomy.

For example, the animator might be

Two desirable objectives which we do not directly address are "naturalness" of the resulting

motion (as opposed to physical realism) and interactivity.

While both of these features are

important to have in an animation system, the problem of generating bipedal locomotion subject to

the above goals is a sufficiently challenging intermediate goal.

Nevertheless, the

proposed

technique affords the animator the freedom to potentially obtain natural looking motions with

reasonable additional effort compared to generating basic course motions.

As well, expected

increases in computer performance over the next year or two promise to make interactive use of

the system a realizable goal.

1. 4

Thesis Organization

This thesis is divided into 6 chapters.

Chapter 2summarizes the previous related work and

presents the background material necessary to understand the chapters which follow.

It also

provides an overview of our animation system.

Chapter 3discusses the underlying principle of

our control approach.

It further describes the general control structure and its application to the

generation of balanced, cyclic locomotion.

Chapter 4presents the basic results of applying the

control formulation to bipedal walking.

Chapter 5 describes further results for variations on

walking gaits. The ability to have the walking biped follow a desired path is also demonstrated.

Finally, Chapter 6 concludes the thesis and discusses a number of possible directions for future

work.