Probabilistic-Based Modeling of Human Perceptual-Motor Behaviors with Application to Wheelchair Targeting Locomotion
Abstract
The objective of this thesis is to develop an empirically-based probabilistic model of the perception-action behaviors of human wheelchair locomotion and navigation over short distance ranges relevant for small space scenarios. More specifically, a probabilistic description and model is sought for the visual perception of distance and orientation with respect to objects, and for wheelchair locomotion based upon straight line and turning behaviors. Further, the specific behavior of manual wheelchair users boarding a small platform area, relevant for lifts and small elevators, is studied experimentally and in a probabilistic-based simulation. The approach taken consists of two parts. First, human subject studies were conducted to characterize the errors of egocentric human spatial perception (i.e., distance and azimuth direction) and locomotion in a near-body 'interactive range', of approximately one to two meters. Second, a predictive model and simulation of locomotion was developed based on data collected in Part I, in order to describe manual wheelchair user locomotion and navigation. The experimental results showed that the main source of perception error was from direction perception. Further, the average wheelchair paths taken in navigation experiments were found to approximate smoothly connected straight line and circular arc (constant turn radius) segments. The simulation results of Part II generally demonstrated good agreement with experimental data based on correlation analysis. Overall, the probabilistic modeling architecture used here enables the simulation of manual wheelchair locomotion and navigation using Hidden Markov Model and Bayes interpretation. The results presented here could inform the development of simulation tools for evaluating the accessibility and mobility of the built environment and assistive devices.
Degree
Ph.D.
Advisors
Seipel, Purdue University.
Subject Area
Engineering|Health sciences|Systems science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.