An energy-efficient learning solution for the Agile Earth Observation Satellite Scheduling Problem

2025-03-03

AI Summary

This paper addresses the Agile Earth Observation Satellite Scheduling Problem (AEOSSP), jointly optimizing target selection, dynamic observation timing, and energy/memory constraints while modeling multi-source image quality degradations—including cloud occlusion, atmospheric turbulence, and resolution degradation. Method: We propose a novel two-stage deep reinforcement learning framework: Stage I generates candidate target sequences; Stage II optimizes precise observation timings under time-varying rewards and coupled resource constraints. Our approach unifies time-varying reward modeling, dynamic environmental state representation, and quality-aware resource scheduling—first of its kind for AEOSSP. Results: Experiments demonstrate a >60% reduction in substandard image rate, 78% decrease in attitude maneuver energy consumption, and significant improvements in mission completion rate and high-quality image acquisition volume.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Learning to Search
• Computer Vision: Remote Sensing / Geospatial AI

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

The Agile Earth Observation Satellite Scheduling Problem (AEOSSP) entails finding the subset of observation targets to be scheduled along the satellite's orbit while meeting operational constraints of time, energy and memory. The problem of deciding what and when to observe is inherently complex, and becomes even more challenging when considering several issues that compromise the quality of the captured images, such as cloud occlusion, atmospheric turbulence, and image resolution. This paper presents a Deep Reinforcement Learning (DRL) approach for addressing the AEOSSP with time-dependent profits, integrating these three factors to optimize the use of energy and memory resources. The proposed method involves a dual decision-making process: selecting the sequence of targets and determining the optimal observation time for each. Our results demonstrate that the proposed algorithm reduces the capture of images that fail to meet quality requirements by>60% and consequently decreases energy waste from attitude maneuvers by up to 78%, all while maintaining strong observation performance.

Problem

• Optimize satellite observation scheduling under energy and memory constraints.
• Address challenges like cloud occlusion and image resolution in satellite imaging.
• Reduce energy waste and improve image quality using Deep Reinforcement Learning.

Innovation

• Deep Reinforcement Learning optimizes satellite scheduling.
• Dual decision-making selects targets and observation times.
• Algorithm reduces energy waste and improves image quality.