EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2997. Satellite Observation Scheduling Problem: an application on Mars Express mission

Invited abstract in session TB-25: Applications of combinatorial optimization II, stream Combinatorial Optimization.

Tuesday, 10:30-12:00
Room: 011 (building: 208)

Authors (first author is the speaker)

1. Benedetta Ferrari
DISMI, University of Modena and Reggio Emilia
2. Maxence Delorme
Tilburg University
3. Manuel Iori
DISMI, University of Modena and Reggio Emilia
4. Marco Lippi
Department of Information Engineering, University of Florence
5. Roberto Orosei
Institute of Radioastronomy, Italian National Institute of Astrophysics

Abstract

In recent years, space exploration has received increasing attention, thus underlining the need for efficient management of space missions. Within this field, Satellite Observation Scheduling Problem (SOSP) concerns the scheduling of observations performed by satellites orbiting around Earth or other celestial bodies. The SOSP has been largely addressed in the last 30 years, dealing with both Earth-centered and outer space missions, with a lot of variants developed to cope with diverse applications and constraints.
Our research focuses on the study of SOSP for the radar MARSIS onboard Mars Express mission, which observes the subsurface of Mars to map the presence of water. Specifically, the Mars Observation Scheduling Problem (MOSP) aims to optimally schedule MARSIS observations to reach a maximum quality coverage of the South Pole of Mars. The MOSP is of high difficulty and has been only manually solved until now.
To solve the MOSP we use an approach combining machine learning and optimization. We first employ a neural network to predict the quality of future observations, starting from a massive historical dataset. We then look for the solution that maximizes the predicted quality by invoking different techniques, including an Integer Linear Program and a set of constructive heuristics and matheuristic algorithms. The resulting algorithm is tested on several real-world instances and scenarios derived from the Mars Express mission, showing good performance.

Keywords

Status: accepted

Back to the list of papers