Experimental Characterization of Multi-Objective Optimization in Human Walking
Open Access
Author:
Starr, Jean
Area of Honors:
Engineering Science
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Joseph Paul Cusumano, Thesis Supervisor Gary L. Gray, Thesis Honors Advisor
Keywords:
cost weights mixture parameter locomotion
Abstract:
Humans walk every single day, yet few understand how human locomotion is achieved. Walking can be broken down into two main objectives: to avoid falls and to change positions while following some path. This means that walking is multi-objective, that is, there can be a range of walking parameters that contribute in harmony to the overall task. These parameters can be analyzed using cost functions which contribute to the study of multi-objective walking. The cost functions show that walking parameters receive a specific weight distribution to maintain locomotion. We hypothesize that maintaining step width, and therefore balance, is more important in locomotion than maintaining lateral body position. Furthermore, we hypothesize that the weight of each cost changes when faced with different walking conditions. This paper explores the importance of step width and lateral body position in human locomotion to determine the cost weights given to these walking parameters and their changing magnitudes to counteract perturbations. Linear regression and significance testing will be used to extract and analyze cost weights in optimal regulation models.
