Prompt-Based PID Tuning for Adaptive AGV Control via In-Context Learning with Few-Shot and MultiSpeed Strategies

This dataset includes PID tuning experiments for an Automated Guided Vehicle (AGV) operating at different speeds (5, 10, 15, 20, and 25 m/min) on a magnetic track. The data, stored in five JSON files, contains detailed logs of PID configurations (Kp, Ki, Kd), performance metrics (RMSE, duration), success indicators, and observations for each run. Additionally, two prompt text files are provided: one uses only multi-speed reference data (prompt_multispeed_learning.txt), while the other combines this with a few-shot+multispeed dataset from the target speed (prompt_multispeed_fewshot_learning.txt). Both prompt strategies were specifically designed by us and used with a Large Language Model (LLM) to predict optimal PID values. This dataset supports the development and evaluation of machine learning-based approaches for speed-dependent PID parameter tuning using our custom transfer learning and few-shot learning frameworks.