Age-related gait motion generation based on biomechanical observations

Abstract

In computer animation, movie, and video game, character’s age is a factor that can change naturally as the story progresses. A lot of studies have conducted to change the character according to age, yet most of them have focused on the facial aging simulation. In this dissertation, a novel approach for generating gait motions according to desired age is presented. The proposed approach transforms the motion based on biomechanical observation. Given an original motion with desired age, our method first detects gait cycles, and then transforms it to achieve the desired posture, gait speed, the duration of double support phase, and peak angle of joints by exploiting the following biomechanical observations. The spine tends to bend, as human gets old. Gait speed decreases linearly according to age. Older people spend more time in the double support phase for balance than younger people. Peak ankle plantarflexion angle, peak hip extension angle, and peak angle of anterior pelvic tilt show statistically significant difference with age. Exploiting these observations, we propose a simple but effective method to generate gait motions automatically according to the desired age. Through experiments, we show that our system can generate natural and realistic gait motions of various ages with a single input motion.