Predictive Dynamics - Human Motion Simulation: Predictive Dynamics - page 94

CHAPTER
5

Predictive Dynamics

Only two things are infinite, the universe and human stupidity, and I'm not
sure about the former.
Albert Einstein (18791955)

5.1 Introduction
Predictive dynamics is a term coined to represent the methodology for predicting
human motion while taking into consideration the biomechanics of the human
and the physics of the task. It is dynamics in the sense that it deals with the equa-
tions of motion. It is predictive in that it is concerned with simulating or calculat-
ing what a human would do under the same conditions. At the heart of predictive
dynamics is the cause and effect.
Predictive dynamics (PD) is an optimization-based approach for human
motion prediction with unknown generalized forces and joint angle profiles. The
basic idea is to model a redundant dynamical system as an optimization problem
and to solve for its motion while optimizing a performance measure and satisfy-
ing the physical and kinematical constraints. Both the motion and the forces that
cause the motion are unknown in the equations of motion that are treated as
equality constraints and evaluated using inverse dynamics instead of their numeri-
cal integration. Available experimental data, such as response at discrete time
points, can be used as constraints in the optimization formulation ( Arora and
Wang, 2005; Kim et al., 2005; Wang et al., 2005; Xiang et al., 2007; Xiang,
2008; Xiang et al., 2010a c ). This chapter is an adaptation of the work by Xiang
et al. (2009c, 2010c) .

5.2 Problem formulation
The basic idea of predictive dynamics is to model a redundant dynamical system
as an optimization problem and to solve for its motion where only limited infor-
mation about the system is available. It uses cost functions, which represent
human performance measures, to drive the motion. The predictive dynamics
approach requires three main components, similar to the basic optimizations