Robotics Research Lab
USC Computer Science
USC Engineering
/ Research / Projects / Evaluating Arm Imitation
Evaluating Arm Imitation
Overview Approach Images and Videos Publications Support Contact Details


This project is in the area of movement imitation. Specifically, it focuses on arm movement imitation. The goal of the project is to provide appropriate feedback and repeated, amplified, or focused demonstrations within the imitation teaching context between the learner and teacher. The framework of the project will permit for either the learner or the teacher to be automated. This gives rise to two possible contexts: a robot teaching a human or a human teaching a robot.


The process of learning arm movements by imitation includes the following steps:

1) teacher demonstrates the task

2) learner imitates

3) teacher evaluates the learner's performance

4) teacher provides instructive feedback in the form of another demonstration, focusing on the least accurate part of the performance, the one to focus on improving first

The third step is very hard, and it is still an open research question. The focus of this project is thus developing a good metric for comparing the movements of the teacher and the learner. The approach to developing this metric is based on segmenting the motion into basic movements. If the segmentation is satisfactory, then time scaling can be performed on each segment, and the error in learning can be isolated to specific segments and joints where the imitation failed.










A. Constantin and M. J Mataric', "Evaluating Arm Movement Imitation",USC Center for Robotics and Embedded Systems (CRES) Technical Report, CRES-05-009, 2005 [PDF].



This work is supported by CRA-W Distributed Mentoring Program, Summer 2005 Award.

The web site for this internship can be found here.

Contact Details

Alexandra Constantin: Alexandra dot E dot Constantin at williams dot edu