Provided by Cognitive Sciences ePrint Archive
Humanoid Motion Description Language
BEN CHOI & YANBING CHEN
Computer Science, College of Engineering and Science
Louisiana Tech University, Ruston, LA 71272, USA
Abstract
In this paper we propose a description language for
specifying motions for humanoid robots and for
allowing humanoid robots to acquire motor skills.
Locomotion greatly increases our ability to interact with
our environments, which in turn increases our mental
abilities. This principle also applies to humanoid robots.
However, there are great difficulties to specify
humanoid motions and to represent motor skills, which
in most cases require four-dimensional space
representations. We propose a representation
framework that includes the following attributes:
motion description layers, egocentric reference system,
progressive quantized refinement, and automatic
constraint satisfaction. We also outline strategies for
acquiring new motor skills by learning from trial and
error, macro approach, and programming. Then, we
outline the development of a new humanoid motion
description language called Cybele.
1. Introduction
Locomotion greatly increases our ability to interact with
our environments, which in turn increases our mental
abilities. The principle that mental abilities can be
improved by interacting with the environments is the
basis for MIT Cog’s project [Brooks 98]. However, Cog
robot currently lacks locomotion. On the other hand,
Honda Humanoid Robots [Honda] possess the state of
the art locomotion system, but lack the autonomy and
the learning abilities. We envision the union of these
two types of robots as the basis of our investigation.
The humanoid robots of the near future will possess the
abilities for locomotion, autonomy, and learning. Much
research remains to be done on such autonomous
humanoid robots. In this paper, we will focus on issues
of developing a common framework for both specifying
motions and for autonomously acquiring motor skills
for such robots.
A unified framework to address both specifying
motions and acquiring motor skills will facilitate the
developments of autonomous humanoid robots. Neural
Network, for example, may be a good medium for
capturing and classifying motor skills. However, the
resultant representation in terms of matrix of weights of
connectivity is difficult to be interpreted and modified.
Thus, in this investigation, we choose to use symbolic
approach by developing a description language.
Our humanoid motion description language, like any
other languages, consists of syntactic and semantic
aspects. Syntactic aspect specifies rules for combining
words while semantic aspect specifies structures for
interpretation and such provides the meaning. We
propose different set of words and rules for different
level of abstraction, such as using joint angles at low
level and using walk and jump at high level of
abstraction. The interpretation and the meaning are
based from our framework that includes egocentric
reference system, progressive quantized refinement, and
automatic constraint satisfaction.
Our language and our framework are unique in many
ways comparing to other related research. Our reference
system simplifies specification of locomotion and
allows motions to be described by uniform and
deterministic expressions. Our concept of Progressive
Quantized Refinement allows a humanoid robot to
interact with its environments using different level of
granularity. Our Automatic Constraint Satisfaction
system reduces the complexity of specifying humanoid
motions. Moreover, our underlining model using non-
deterministic finite state machines allows humanoid
robots to learn new motor skills.
Research in describing humanoid motions begins with
the works for describing human dances. Popular dance
notation systems include Benesh [Causley 80],
Labanotation [Hutchinson 87], and EW [EW]. Benesh
is the simplest one and is designed particularly for
dance description. Labanotation is more comprehensive
for describing human motion in general. EW can be
applied on linkage systems other than human body.
Computers are now used to aid the interpretation and
visualization of these notations [Singh 84] [Adamson
87] [Calvert 93] [Schiphorst 92]. Researchers used