Personal and Emotive Walking Styles For a Physical Robot By Emulating a Human

In this project the aim is to extract personal and emotive walking patterns from a video dataset of individuals walking, and to then emulate these on a robotic walking device, so the robot becomes an abstracted form of the most characteristic walking patterns of an individual -similar to an animated caricature. In this way a robot can borrow a personal way of walking, and also vary this individualized walking style to reflect various emotional states. The robot's movements should first be simulated in a virtual environment and then transferred to a physical robot.

The goal of this project: A personal walking pattern that includes things like a minor limp or other physical conditions of the individual. Examples of emotional variables could be Depressed - Happy, Self-conscious - Confident, Confused - Directional, Shy - Curious. There can also be variables like Tired - Energetic. The objective is alsot buils a designed behavior interface that allows one to give high level mood commands, that is then reflected in the walking style of the robot.

Supervised by Ce Zhang

ce.zhang@inf.ethz.ch