Robots learn to plan and adapt in real time with BrainBody-LLM AI tech

Researchers at NYU Tandon School of Engineering have developed BrainBody-LLM, an innovative AI algorithm that enables robots to plan, adapt, and learn in real time, mimicking human-like movement and decision-making. Unlike traditional robotic systems or existing large language model (LLM)-based planners that often produce ungrounded or inflexible plans, BrainBody-LLM integrates two LLM components: the "Brain" for high-level task planning and the "Body" for translating plans into precise actuator commands. A key advancement is its closed-loop feedback system, allowing the robot to continuously monitor its actions and environment, sending error signals back to the LLMs to dynamically adjust and correct movements during task execution. Testing on both virtual simulations (VirtualHome) and a physical robotic arm (Franka Research 3) demonstrated BrainBody-LLM’s superior performance, with up to a 17% increase in task completion rates in simulations and successful handling of real-world complexities on the physical robot.