arXiv subjects

Samuel Balula

Publications and source records attributed to Samuel Balula.

4 recordsLinked to original sources

arXiv2023

Sequential Quadratic Programming-based Iterative Learning Control for Nonlinear Systems

Learning-based control methods for industrial processes leverage the repetitive nature of the underlying process to learn optimal inputs for the system. While many works focus on linear systems, real-world problems involve nonlinear dynamics. In this work, we propose an algorithm for the nonlinear iterative learning control problem based on sequential quadratic programming, a well-studied method for nonconvex optimization. We repeatedly solve quadratic subproblems built using approximate nonlinear models and process measurements, to find an optimal input for the original system. We demonstrate our method in a trajectory optimization problem for a precision motion system. We present simulations to illustrate the performance of the proposed method for linear and nonlinear dynamics models.

eess.SY↗

arXiv2022

Drone-based Volume Estimation in Indoor Environments

Volume estimation in large indoor spaces is an important challenge in robotic inspection of industrial warehouses. We propose an approach for volume estimation for autonomous systems using visual features for indoor localization and surface reconstruction from 2D-LiDAR measurements. A Gaussian Process-based model incorporates information collected from measurements given statistical prior information about the terrain, from which the volume estimate is computed. Our algorithm finds feasible trajectories which minimize the uncertainty of the volume estimate. We show results in simulation for the surface reconstruction and volume estimate of topographic data.

eess.SY↗

arXiv2022

Data-driven Reference Trajectory Optimization for Precision Motion Systems

We propose a data-driven optimization-based pre-compensation method to improve the contour tracking performance of precision motion stages by modifying the reference trajectory and without modifying any built-in low-level controllers. The position of the precision motion stage is predicted with data-driven models, a linear low-fidelity model is used to optimize traversal time, by changing the path velocity and acceleration profiles then a non-linear high-fidelity model is used to refine the previously found time-optimal solution. We experimentally demonstrate that the proposed method is capable of simultaneously improving the productivity and accuracy of a high precision motion stage. Given the data-based nature of the models, the proposed method can easily be adapted to a wide family of precision motion systems.

eess.SY↗

arXiv2020

Reference design for closed loop system optimization

An optimization-based method for improving the productivity of precision machine tools is proposed, where the reference path is computed in local coordinates, and information about the machine tool performance is learned from experimental data. The optimization yields a modified reference that is tracked by the existing low-level controller. The method is tested in simulation for a biaxial positioning system. The positioning system is modelled as double integrator, and the controller characteristic is modelled from experimental data using a least-squares fit. Simulation results show that the method is effective in designing optimal references even for challenging geometries such as sharp corners. The application of this procedure allows the retrofit of the control of existing machines with minimal overhead, by providing a modified reference file to track.

eess.SY↗