This study explores a novel approach of using large language models (LLMs) in the real-time Proportional–Integral–Derivative (PID) control of a physical system, the Quanser QUBE-Servo 2. We investigated whether LLMs, used with an Artificial Intelligence (AI) agent workflow platform, can participate in the live tuning of PID parameters through natural language instructions. Two AI agents were developed: a control agent that monitors the system performance and decides if tuning is necessary, and an Optimizer Agent that updates PID gains using either a guided system prompt or a self-directed free approach within a safe parameter range. The LLM integration was implemented through Python programming and Flask-based communication between the AI agents and the hardware system. Experimental results show that LLM-based tuning approaches can effectively reduce standard error metrics, such as IAE, ISE, MSE, and RMSE. This study presents one of the first implementations of real-time PID tuning powered by LLMs, and it has the potential to become a novel alternative to classical control, as well as machine learning or reinforcement learning-based approaches. The results are promising for using agentic AI in heuristic-based tuning and the control of complex physical systems, marking the shift toward more human-centered, explainable, and adaptive control engineering.

The interactions between body tissues and a focused ultrasound beam can be evaluated using various numerical models. Among these, the Rayleigh–Sommerfeld and angular spectrum methods are considered to be the most effective in terms of accuracy. However, they are computationally expensive, which is one of the underlying issues of most computational models. Typically, evaluations using these models require a significant amount of time (hours to days) if realistic scenarios such as tissue inhomogeneity or non-linearity are considered. This study aims to address this issue by developing a rapid estimation model for ultrasound therapy using a machine learning algorithm. Several machine learning models were trained on a very-large dataset (19,227 simulations), and the performance of these models were evaluated with metrics such as Root Mean Squared Error (RMSE), R-squared (R²), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). The resulted random forest provides superior accuracy with an R² value of 0.997, an RMSE of 0.0123, an AIC of −82.56, and a BIC of −81.65 on an external test dataset. The results indicate the efficacy of the random forest-based model for the focused ultrasound response, and practical adoption of this approach will improve the therapeutic planning process by minimizing simulation time.

Deep Learning for Engineers introduces the fundamental principles of deep learning along with an explanation of the basic elements required for understanding and applying deep learning models. As a comprehensive guideline for applying deep learning models in practical settings, this book features an easy-to-understand coding structure using Python and PyTorch with an in-depth explanation of four typical deep learning case studies on image classification, object detection, semantic segmentation, and image captioning. The fundamentals of convolutional neural network (CNN) and recurrent neural network (RNN) architectures and their practical implementations in science and engineering are also discussed. This book includes exercise problems for all case studies focusing on various fine-tuning approaches in deep learning. Science and engineering students at both undergraduate and graduate levels, academic researchers, and industry professionals will find the contents useful. <!----> <!----> <!----> <!---->

This study explores a novel approach of using large language models (LLMs) in real-time Proportional-Integral-Derivative (PID) control of a physical system, the Quanser QUBE-Servo 2. We investigated whether LLMs, used with an Artificial Intelligence (AI) agent workflow platform, can participate in live tuning of PID parameters through natural language instructions. Two AI agents were developed: a control agent that monitors the system performance and decides if tuning is needed, and an optimizer agent that updates PID gains using either a guided system prompt or a self-directed free approach within a safe parameter range. The LLM integration was implemented through Python programming and Flask-based communication between the AI agents and the hardware system. Experimental results show that both tuning approaches effectively reduced standard error metrics, such as IAE, ISE, MSE, and RMSE. This study presents the first known real-time implementation of servo motor control powered by LLMs, and it has the potential to become a novel alternative to classical control or machine learning and reinforcement learning based control approaches. The results are promising for using agentic AI in heuristic-based tuning and control of complex physical systems.