Incremental learning of humanoid robot behavior from natural interaction and large language models

• Published in: Frontiers in robotics and AI 2024-10, Vol.11, p.1455375
• Main Authors: Bärmann, Leonard, Kartmann, Rainer, Peller-Konrad, Fabian, Niehues, Jan, Waibel, Alex, Asfour, Tamim
• Format: Article
• Language: English
• Subjects: cognitive modeling; humanoid robots; human–robot interaction; incremental learning; knowledge representation for robots; large language models; Robotics and AI
• ISSN: 2296-9144; 2296-9144
• DOI: 10.3389/frobt.2024.1455375

Natural-language dialog is key for an intuitive human-robot interaction. It can be used not only to express humans' intents but also to communicate instructions for improvement if a robot does not understand a command correctly. It is of great importance to let robots learn from such interaction experiences in an incremental way to allow them to improve their behaviors or avoid mistakes in the future. In this paper, we propose a system to achieve such incremental learning of complex high-level behavior from natural interaction and demonstrate its implementation on a humanoid robot. Our system deploys large language models (LLMs) for high-level orchestration of the robot's behavior based on the idea of enabling the LLM to generate Python statements in an interactive console to invoke both robot perception and action. Human instructions, environment observations, and execution results are fed back to the LLM, thus informing the generation of the next statement. Since an LLM can misunderstand (potentially ambiguous) user instructions, we introduce incremental learning from the interaction, which enables the system to learn from its mistakes. For that purpose, the LLM can call another LLM responsible for code-level improvements in the current interaction based on human feedback. Subsequently, we store the improved interaction in the robot's memory so that it can later be retrieved on semantically similar requests. We integrate the system in the robot cognitive architecture of the humanoid robot ARMAR-6 and evaluate our methods both quantitatively (in simulation) and qualitatively (in simulation and real-world) by demonstrating generalized incrementally learned knowledge.