Technical Papers

Biped Control

Thursday, 29 July | 3:45 PM - 5:15 PM | Room 502 B
Session Chair: Cindy Grimm, Washington University in St. Louis
Sampling-Based Contact-Rich Motion Control

Given a motion trajectory, this method reconstructs its control by randomized sampling. The paper demonstrates fast reconstruction for a diverse set of captured motions, including contact-rich tasks such as forward and sideways rolls. Physically plausible motion transformation and retargeting, and reference-trajectory-free idling can be done within the same framework.

Libin Liu
Tsinghua University

KangKang Yin
Microsoft Research Asia

Michiel van de Panne
The University of British Columbia

Tianjia Shao
Tsinghua Univeristy

Weiwei Xu
Microsoft Research Asia

Data-Driven Biped Control

A dynamic controller to physically simulate three-dimensional full-body biped locomotion. The data-driven controller uses motion capture reference data to reproduce realistic human locomotion through real-time simulation. This paper demonstrates the effectiveness of the approach through examples that allow bipeds to turn, spin, and walk while their direction is steered interactively.

Yoonsang Lee
Seoul National University

Sungeun Kim
Seoul National University

Jehee Lee
Seoul National University

Generalized Biped Walking Control

A simple control strategy for physically simulated walking that generalizes well across gait parameters, motion styles, character proportions, and a variety of skills, such as picking up objects placed at any height, pushing and pulling, stepping over obstacles, and walking with crates.

Stelian Coros
The University of British Columbia

Philippe Beaudoin
The University of British Columbia

Michiel van de Panne
The University of British Columbia

Feature-Based Locomotion Controllers

A technique for building locomotion controllers for physics-based characters in terms of high-level features. Objective terms control each feature and are combined using a strict prioritization algorithm. Using this approach, human-like qualities emerge automatically and control can be mapped onto new bipeds without modifications.

Martin de Lasa
University of Toronto

Igor Mordatch
University of Toronto

Aaron Hertzmann
University of Toronto

In the News
SIGGRAPH 2010 Video