Continual Robot Skill and Task Learning via Dialogue

📰 ArXiv cs.AI

Robots learn new skills and tasks through dialogue interactions with humans using a framework that utilizes a large language model (LLM) and a skill library

advanced Published 31 Mar 2026

Action Steps

Maintain a skill library to store learned skills and tasks
Utilize an existing LLM to generate and understand dialogue interactions
Implement a dialogue system for the robot to query humans for novel skills
Continually update the skill library with new skills and tasks learned through dialogue interactions

Who Needs to Know This

Robotics engineers and AI researchers can benefit from this framework as it enables robots to learn novel skills and tasks efficiently, while human-robot interaction designers can improve the dialogue system for more effective communication

Key Insight

💡 Continual learning of robot skills and tasks can be achieved through human-robot dialogue interactions

Key Takeaways

Robots learn new skills and tasks through dialogue interactions with humans using a framework that utilizes a large language model (LLM) and a skill library

Full Article

Title: Continual Robot Skill and Task Learning via Dialogue

Abstract:
arXiv:2409.03166v3 Announce Type: replace-cross Abstract: Interactive robot learning is a challenging problem as the robot is present with human users who expect the robot to learn novel skills to solve novel tasks perpetually with sample efficiency. In this work we present a framework for robots to continually learn tasks and visuo-motor skills and query for novel skills via dialog interactions with human users. Our robot agent maintains a skill library, and uses an existing LLM to perform grou

Read full paper → ← Back to Reads