Model validation for a pair of biologically abstracted quadruped forelimbs through the design of experiment (DOE) technique

Abstract

This paper presents a statistical approach used for model validation of a pair of biologically abstracted quadruped forelimb. The forelimb design is abstracted from the skeletal anatomy of a typical canine dog, introducing an additional link in place of the scapula. Parameter limits are obtained from known biological data of a typical canine during locomotion. Model performance is based on quadruped forelimb trot gaits at 0.8, 1 and 1.2 Hz stride frequencies with sinusoidal joint rotations. Successive applications of the design of experiment statistical tool allows for quick model validation with respect to model variables. Results are confirmed using tedious factorial simulations coupled with a simple two-stage manual verification procedure. Forelimb performance indicates a correlation of high-speed trot motion with narrow profiles with model stability dominated by the neck-spine joint at low speeds. © 2010 The Institute of Measurement and Control.