Behavior Tree
Behavior trees (BTs) are hierarchical, modular frameworks for representing and executing complex robot behaviors, aiming to improve the design, readability, and adaptability of robot control systems. Current research emphasizes the integration of large language models (LLMs) to automate BT generation from natural language instructions or demonstrations, enhancing efficiency and enabling easier task programming for non-experts. This approach is proving valuable in diverse applications, including autonomous vehicles, robotic manipulation, and human-robot collaboration, by facilitating more robust, adaptable, and verifiable robot control policies.
Papers
CoBT: Collaborative Programming of Behaviour Trees from One Demonstration for Robot Manipulation
Aayush Jain, Philip Long, Valeria Villani, John D. Kelleher, Maria Chiara Leva
LLM-BT: Performing Robotic Adaptive Tasks based on Large Language Models and Behavior Trees
Haotian Zhou, Yunhan Lin, Longwu Yan, Jihong Zhu, Huasong Min