Near-Optimal Covering Solution for USV Coastal Monitoring using PAES - Ecole Nationale d'Ingénieurs de Brest Accéder directement au contenu
Article Dans Une Revue Journal of Intelligent and Robotic Systems Année : 2022

Near-Optimal Covering Solution for USV Coastal Monitoring using PAES

Résumé

This paper addresses a multi-objective optimization problem for marine monitoring using USV. The objectives are to 1) cover the maximum area with 2) the lowest energy cost while avoiding collisions. To solve this problem, we use two optimization approaches: The first one is an exact method based on a Mixed Integer Programming model of the Covering Salesman Problem (CSP) and Travelling Salesman Problem with Profit (TSPP). The second one is a heuristic method based on a multi-objective evolutionary algorithm known as Pareto Archived Evolution Strategy (PAES). Then, a customized PAES (C-PAES) is proposed where the optimal chromosome size is calculated to enhance both performance and solutions quality. To validate the proposed heuristic solution, C-PAES is compared with the exact method. The results show that if the monitored area is represented with a graph containing less than 74 vertices, C-PAES is able to provide optimal solution within a much lower execution time. Beyond the limit of 74 vertices, the exact solver is not able to provide any solution within 24 hours delay while C-PAES can converge to a solution which consumes only 7% additional energy as comped to the optimal one.
Fichier principal
Vignette du fichier
00JINT_revised_paper.pdf (2.2 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03780408 , version 1 (27-09-2022)

Identifiants

Citer

Hand Ouelmokhtar, Yahia Benmoussa, Jean-Philippe Diguet, Djamel Benazzouz, Laurent Lemarchand. Near-Optimal Covering Solution for USV Coastal Monitoring using PAES. Journal of Intelligent and Robotic Systems, 2022, 106 (1), pp.24. ⟨10.1007/s10846-022-01717-x⟩. ⟨hal-03780408⟩
97 Consultations
59 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More