Skip to main content
eScholarship
Open Access Publications from the University of California

UC Riverside

UC Riverside Electronic Theses and Dissertations bannerUC Riverside

Simulation and Control of Humanoid Rolling and Falling Behaviors

Abstract

Controlling physically based characters is an ongoing

problem in animation. Motions that involve a large of

changing contacts with the environment are especially

problematic. In this thesis we present techniques for

controlling contact rich motions, specifically rolling and

falling motions. For rolling we present a method for

controlling the full body orientation of a multibody

character. We use this to describe a roll for the character

to perform in terms of the character's goal orientation,

linear velocity and angular velocity. We then use

a quadratic programming controller to compute the torques to

best achieve these goals. To deal with the many and

constantly varying contacts we introduce a heuristic that

selects only the most important contact points to be active

in the quadratic program. Falling is similar except it

requires time varying orientation and position goals for the

character. We automatically compute these goals by defining

a function to evaluate the quality of a fall and use

sampling based optimization to find the best way to control

the character for a given fall. We demonstrate our rolling

controller for several different rolls in different

environments and our falling controller on a variety of

different falls.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View