Concluding Remarks - Human Motion Simulation: Predictive Dynamics - page 234

and fatigue capability. All of these characteristics are input parameters in the PD
environment.
10.1.2 Predictive dynamics solves dynamics without integration
This is perhaps the most significant benefit of PD. Multi-body dynamics typically
attempts to use numerical integration to determine the motion of a system of rigid
bodies that has been subjected to external forces or is in motion. However, multi-
body dynamics methods cannot address the prediction of human motion for the
following reasons:
a. Multi-body dynamics cannot handle high-DOF models having equality and
inequality constraints on the motion, such as those of the human body.
b. The algebraic-differential equations that are formed by the human system are
too complex and are highly nonlinear.
c. Unlike PD, multi-body dynamics does not consider the natural behavior of the
motion. The PD optimization formulation, on the other hand, uses objective
functions to drive the motion.
10.1.3 Predictive dynamics renders natural motion
The essence of the PD method is its ability to render natural motions, similar to
those of humans. It considers the human body to have a set of cost functions
that drive the motion. This is an essential difference from traditional inverse
kinematics methods, which yield postures and motions that cannot be performed
realistically by humans. Predictive dynamics exhibits natural motion because
the motion is constrained by the joint range of motion for all joints, the laws of
physics, and any physical constraints that may exist. More importantly, PD
attempts to find solutions that obey the dynamic strength surface for each joint.
The result is natural motion that looks realistic for a person of a certain stature
and strength.
10.1.4 Predictive dynamics induces natural behavior
The use of one or more cost functions, depending on the task, has been shown to
yield different behaviors. It was also shown that using multi-objective optimiza-
tion can induce certain behaviors typically performed by humans. Whether it is
energy, jerk, discomfort, effort, or torque, it is obvious that humans try to mini-
mize or maximize some quantity. This, we believe, also develops as humans gain
experience, grow wiser, and become adept at performing various tasks.
10.1.5 Predictive dynamics admits cause and effect
A human simulator can be tried, loaded, and pushed to obtain a reaction. The
user of a human simulator that is developed using PD can experience cause and