Model-free active balancing for humanoid robots

Sara McGrath*, John Anderson, Jacky Baltes

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)


To be practical, humanoid robots must be able to manoeuvre over a variety of flat and uneven terrains, at different speeds and with varying gaits and motions. This paper describes three balancing-reflex algorithms (threshold control, PID control, and hybrid control) that were implemented on a real 8 DOF small humanoid robot equipped with a two-axis accelerometer sensor to study the capabilities and limitations of various balancing algorithms when combined with a single sensor. We term this approach a model-free approach, since it does not require a mathematical model of the underlying robot. Instead the controller attempts to recreate successful previous motions (so-called baseline motions). In our extensive tests, the basic threshold algorithm proves the most effective overall. All algorithms are able to balance for simple tasks, but as the balancing required becomes more complex (e.g. controlling multiple joints over uneven terrain), the need for more sophisticated algorithms becomes apparent.

Original languageEnglish
Title of host publicationRoboCup 2008
Subtitle of host publicationRobot Soccer World Cup XII
Number of pages12
Publication statusPublished - 2009
Externally publishedYes
Event12th annual RoboCup International Symposium, RoboCup 2008 - Suzhou, China
Duration: 2008 Jul 152008 Jul 18

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5399 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Other12th annual RoboCup International Symposium, RoboCup 2008

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science


Dive into the research topics of 'Model-free active balancing for humanoid robots'. Together they form a unique fingerprint.

Cite this