Interactive Learning of State Representation through Natural Language Instruction and Explanation

TL;DR

This paper proposes a method for robots to learn new state representations through natural language communication with humans, moving beyond the assumption that robots have complete world models.

Contribution

It introduces an approach for robots to acquire and update state representations interactively via natural language instructions and explanations.

Findings

Enables robots to learn new state predicates through language

Addresses limitations of closed-world assumptions in robot learning

Facilitates adaptive and scalable robot understanding

Abstract

One significant simplification in most previous work on robot learning is the closed-world assumption where the robot is assumed to know ahead of time a complete set of predicates describing the state of the physical world. However, robots are not likely to have a complete model of the world especially when learning a new task. To address this problem, this extended abstract gives a brief introduction to our on-going work that aims to enable the robot to acquire new state representations through language communication with humans.