Enhancing Robotic System Robustness via Lyapunov Exponent-Based Optimization

G. Fadini, S. Coros · 2024

Read Enhancing Robotic System Robustness via Lyapunov Exponent-Based Optimization by G. Fadini, S. Coros (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-06776.txt.

Description

Fadini and Coros wrote this 2024 robotics paper on measuring and then maximizing how gracefully the motion of a robot resists perturbation. Stability is usually judged qualitatively, yet design optimization needs a number. Their metric is the Lyapunov exponent spectrum, computed through differentiable simulation run over long horizons, which naturally extends to the limit cycles of walking and other rhythmic tasks.

A gradient-based optimizer built on JAX then tunes designs and controllers against the exponent objective. The test bench spans simple systems up to high-degree-of-freedom machines in contact-rich settings where parts collide and grip. Across the board the optimization stiffened robustness in ways cruder metrics miss. The authors pitch exponent-guided design as a general tool for robots that must not fall over, in simulation or on the floor.

Catalog record

CREATORG. Fadini, S. Coros
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-06776

Browse the full library · About the Tower of Babel