Paper1: Robust Physics-Based Locomotion Using Low-Dimensional Planning

by csv on January 23, 2011

in Assignment 1

1. Sentence: ( 2nd paragraph: 1st line)
This paper introduces a locomotion controller for full-body 3D physics-based  characters that plans joint torques by optimizing a low
dimensional physical model.

2. Problem: ( 1st paragraph )
Full space-time motion planning on irregular terrains with zero error
tolerance is computational extremely expensive and would not be able to handle unforeseen disturbances, changes to the environment, or new user-specified goals. Existing simplified dynamics models are gait specific and do not plan control over multiple foot-steps.

3. Key Idea:
At each instant, the controller optimizes a plan over multiple phases
of locomotion. To generate the plan, full-body dynamics are approximated by a set of closed form equations of motion, allowing re-planning to occur at every time step. The controller then chooses optimal full-body torques according to the current plan, while maintaining balance.

4. What this paper lets us do:
This work allows navigation over uneven terrain, including gaps, inclines, stairs, large drops and stepping stones. This method does not require any motion capture data and is fairly robust to external disturbances.

5. Additional resources:
Both SIGGRAPH 2010 presentation slides and video available at:
http://www.dgp.toronto.edu/~mdelasa/slip

Previous post:

Next post: