Jorge Vicente

In this work, we present a time-optimal trajectory generation method for 3D overhead cranes using differential flatness. Our approach incorporates complex physical constraints like nonlinear friction and collision avoidance for both the payload and rope. We demonstrate that modeling friction is crucial for avoiding actuator saturation and collisions, ensuring fast, safe, and reliable crane operations.

This paper presents a new approach to accurately simulating 3D overhead cranes with friction. Although nonlinear friction dynamics has a significant impact on these systems, accurately modeling this phenomenon in simulations is a significant challenge. Furthermore, we present a step-by-step algorithm for the comprehensive estimation of all unknown system parameters, including friction. This methodology is based on Bayesian Linear Regression and Least Squares (LS) estimations.

This work introduces a neural network-based approach for the rapid and automated estimation of parameters in nonlinear physical systems. It addresses the challenge of manufacturing variability in mass-produced low-cost devices, where physical tolerances can lead to significant differences between individual units, complicating modeling and control. The proposed method uses convolutional layers to process temporal signals, extracting relevant features, and dense layers to calculate the system's parameters. This approach enables efficient parameter estimation from a single experiment, offering advantages in industrial settings where speed and automation are critical.

This work tackles the challenge of precisely controlling fluid flow in microfluidic systems, essential tools in diverse scientific domains. It presents a model predictive controller (MPC) for accurate regulation and addresses the need for adaptability by exploring techniques to estimate key system parameters. This allows the MPC to adjust to changes, such as different fluids, with the research demonstrating system identifiability and implementing noise mitigation, validated experimentally to confirm effective parameter estimation and controller updates.