Abstract

Imitating Operations On Internal Cognitive Structures for Language Aquisition

Highlights

  • A robot that is to operate in human populated environments, such as a home or an office, must be able to take the social context into account and take simple directions

  • Therefor the hand sign requests an operation on an internal cognitive structure, and this operation must be imitated even if it is not directly visible

  • In the experiment presented in this paper the syntax of the sign language that the imitator must find is as follows: the first sign requests the internal operation of a specific object focus; a “1” requests focus on the red object, a “2” the green and a “3” the blue object

Read more

Summary

INTRODUCTION

A robot that is to operate in human populated environments, such as a home or an office, must be able to take the social context into account and take simple directions. Experiments show that the model generalizes well to combinations of communicative signs that was not observed during demonstrations Both actions and gestures are continuous, never exactly the same, and the imitator is never presented with any form of symbolic representation of interactant hand signs or demonstrator actions. The proposed algorithm does not include a separate language system but is instead an imitation learning system whose context has been extended to include the communicative hand signs of an interactant. There are difficult practical problems involved with imitating such operations but, as the present paper shows, this is possible if the demonstrator acts according to a consistent policy, the internal state is changed in response to observable parts of the context (in this case the signs of the interactant) and the internal state modifies observable behavior. In a single demonstration there is no information about what the internal state is

ALGORITHMS
Demonstrations
The grouping algorithm
Finding the movement and the coordinate system during reproduction
Reproduction
Simulating the setup
CONCLUSIONS AND FUTURE WORK
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call