" A pioneering research community "

Research Mentor(s)

Professor Maja Mataric
University of Southern California
Professor, Department of Computer Science
Visiting Research Mentor
This email address is being protected from spambots. You need JavaScript enabled to view it.

Dr. Sanza T. Kazadi
Jisan Research Institute
JRI Research Mentor
This email address is being protected from spambots. You need JavaScript enabled to view it.


Introduction and Motivation

Human motor control is one of the most complex processes in the brain, and it is currently poorly understood. Human movement has historically been studied through observation (in psychology), and more recently through taking accurate measurements of the limb positions (in psychometrics), in order to guide theories of motor control. While methods for data gathering have become quite sophisticated, approaches to evaluating the resulting data have not. Our work focuses on learning new skills through imitation, a complex process that is rarely found in nature, aside from a few primates and dolphins. While simple mimicry is not rare (e.g., bird song imitation, "parroting", etc.), acquisition of a completely new skill by observation, considered "true imitation", is based on more complex and thus rarer neural processing.

One of the integral parts of studying imitation is evaluation of imitated movement. On one hand, it is important for the learner to evaluate the quality of what was done in order to correct and further improve its skill. On the other hand, it is imperative for a scientist studying imitation to have objective evaluation metrics for comparing the teacher's movement with that of the imitator.

This project aims at developing a set of evaluation metrics for comparing and scoring human movement based on psychometric data.

Proposed Research

We propose to use a large corpus of human movement data, gathered with a collection of methods:

  • OptoTrak
  • FastTrak

visual systems.

In all cases, the data provides position information about a specific part of the body. Given a human kinematic model (i.e., understanding of the connections between the parts of the limbs and the degrees of freedom of the joints), this data can be used to reconstruct the movement in a simulation (ranging from stick figures to fully dynamic humanoid models) in order to recreate the imitator's performance. This, however, only helps visualization and observational evaluation. What is needed is a set of methods for comparing the analytical description of the teacher's and imitator's movement, in order to score the imitator's performance.

Scoring performance is a highly challenging problem, as it is multifaceted. Given the known limitation of the human perceptual system, we do not expect the imitated movement to faithfully reproduce any geometric patters executed by the end-point (usually the hand), or replicate the precise joint angle values of the teacher. However, we do expect the resulting movement to be similar to the teacher's, and analytically comparable to other more or less similar attempts.


A. Hsieh, S. Kazadi, E. Acevedo, P. Hung, M. Mataric A correlation-based movement comparison measure. Technical Paper of the Jisan Research Institute, 2001.

The Jisan Research Institute · A California 501(c)3 Nonprofit Corporation · © Copyright 2009-2014 · All rights reserved