A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) demonstrates that large language models (LLMs) can significantly accelerate the process of robot motion planning, a traditionally computationally intensive task. The research introduces a method where an LLM, such as GPT-4, breaks down a high-level task into smaller, manageable sub-goals described in natural language. A second, smaller AI model then translates these language-based steps into low-level robot actions. This hierarchical approach, tested on a robotic arm performing tasks like sorting blocks by color, reduced planning time by nearly 50% compared to state-of-the-art traditional planners. The method shows promise for making robots more efficient and adaptable in complex, real-world environments like warehouses or homes. Read the full article at: https://technologyreview.com/2024/07/12/1094756/ai-helps-robots-move-faster/