Predicting the Biomechanics of Walking - Human Motion Simulation: Predictive Dynamics - page 147

CHAPTER
7

Predicting the Biomechanics
of Walking

With contribution by Yujiang Xiang, PhD, Virtual Soldier Research Program

If everything seems under control, you're just not going fast enough.
Mario Andretti

7.1 Introduction
The objective of this chapter is to develop a broadly applicable formulation for
the prediction of human walking. It is to answer the question: “How would a per-
son walk given a set of conditions, such as a specific body size, a specific load,
and joint ranges of motion?”
Past research has addressed walking analysis, also called gait analysis, where
a person's walk is measured and sometimes mathematically modeled for the rea-
sons of analysis, to better understand deviation from the normal or to evaluate
performance. This is of importance for the understanding of disabilities, injuries,
and post-surgical gait, for example. The intent of this chapter is to create a predic-
tive model for the biomechanics of walking.
Walking is also called ambulation and is one of the main gaits of locomotion
among legged animals. Walking is generally defined by an inverted pendulum gait in
which the body vaults over the stiff limb or limbs with each step. This applies regard-
less of the number of limbs—even for arthropods with six, eight or more limbs.
There are many reports that have established methods to simulate human
motion, in particular walking, using databases generated from experiments on
human subjects ( Choi et al., 2003; Pettre and Laumond, 2006 ). These methods
draw upon libraries of pre-stored motions to present the best-fit approach for a
particular scenario. These approaches are limited by the accuracy and amount of
available experimental data.
Other methods have attempted to model and simulate human walking. One
such approach is to solve the walking problem based on the idea that biped walk-
ing can be treated as an inverted pendulum. Advantages of this method are its
simplicity and faster solvable dynamics equations ( Park and Kim, 1998 ).
However, the method also suffers from an inadequate dynamics model that cannot
generate natural and realistic human motion; particularly problematic is its inabil-
ity to represent a large DOF system with high fidelity.