Computer Simulation of Human Movemfnts

Abstract

While a great deal is known about the physiological aspects of the human neuro-musculo-skeletal system, the precise mechanisms whereby individual joint motions are coordinated by neural networks to achieve purposeful motions are rather poorly understood.^1,2 That is, mathematical models capable of explaining the observed behavior of man and higher animals during locomotion and while manipulating objects are in a rather primitive state of development. The author feels that this lack of understanding is due to three primary factors as follows:

• 1. From a mathematical or engineering point of view, the human musculo-skeletal system involves great dynamical complexity, even when it is idealized to a system of rigid levers with simple torque generators acting at each joint.^3,4

• 2. Even greatly simplified differential equation models for human motion are highly nonlinear and cannot be solved by analytic means. There was therefore little point in deriving such models prior to the introduction of efficient means for their numerical solution in the form of modern digital computers.

• 3. Although computers capable of solving very complicated equations of motion are now available, measurement techniques necessary for testing such models against physical observations have been inadequate. Instruments and recording systems capable of automatically determining limb segment motions with the necessary precision have begun to appear only very recently.^5,6,7