Πανεπιστήμιο Μακεδονίας
Optimisation and computational optimal control via local pursuit
Abstract
dc:descriptionThis thesis develops a family of algorithms for solving a class of non-autonomous optimal control problems in which the running cost and the set of admissible states vary with time. Such problems arise when physical systems operate in non-stationary dynamic environments which contain moving, deforming obstacles and whose geometry changes over time. When sensor noise is present in such settings, no ‘off-the-shelf’ method is available; in the noise-free case, the only known approach presumes complete knowledge of the environment’s future evolution. We introduce methods grounded in the idea of local pursuit, inspired by behaviors observed in biological systems such as ants foraging. We adopt a multi-agent formulation in which agents iteratively solve a sequence of optimisation problems by ‘pursuing’ one another, relying exclusively on local information and without imposing restrictive structural assumptions on their dynamics, constraints, or cost functions. This approach results in three algorithms, all of which generate control inputs that produce a trajectory guaranteed to be locally optimal, while respecting state and input constraints. The algorithms form a progression of increasing generality in terms of the environments they can be applied in. We first extend prior local pursuit approaches from static environments to noise-free non-stationary ones, preserving convergence guarantees and matching the quality of the ‘full-knowledge’ benchmark approach at substantially lower computational cost. Subsequently, we account for noise in both agent and obstacle observations, where convergence guarantees are retained under mild problem-specific conditions. Finally, we restore convergence guarantees where those conditions are violated, at a tractable computational cost. The performance of all three algorithms is demonstrated in extensive numerical simulations.
Degree
thesis:*- Grantor dc:publisher
- Πανεπιστήμιο Μακεδονίας
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Kartsiotis, Georgios
- Καρτσιώτης, Γεώργιος
Subjects
dc:subject × 16- Βέλτιστος έλεγχος
- Επαναληπτική βελτιστοποίηση
- Δυναμικά περιβάλλοντα υπό την παρουσία θορύβου
- Βιοεμπνευσμένες μέθοδοι βελτιστοποίησης
- Τοπική καταδίωξη
- Optimal control
- Iterative optimisation
- Noisy dynamic environments
- Bio-inspired optimisation
- Local pursuit
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Επιστήμη ηλεκτρονικών υπολογιστών, θεωρία και μέθοδοι
- Computer and Information Sciences
- Natural Sciences
- Computer science, theory and methods
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/62833