Imitation of gestures by robotic arms: A quality assessment approach

Abstract

This article proposes 
quantitative evaluation metrics 
of imitation, suited for the 
imitation of four gestures 
performed by a robotic arm on 
a simulation environment; for 
each gesture, the metric relies 
on a group of simple geometric 
shapes. This approach broadens 
the concept of imitation; 
since common metrics just 
accumulate the error between 
the joints of the demonstrator 
and the imitator along their 
trajectories, in other words, 
check for the exact replica of 
the gesture performed by the 
demonstrator. The results of the 
proposed metric are compared 
with an evaluation survey 
conducted in human observers.