Large Language Model-Assisted Adaptive Heuristic Search for Agile Earth Observation Satellite Scheduling

He Wang; Yeye Liu; Ruijie Li; Xuemei Jiang; Junyu Wu; Lining Liu Zhuoqi Man; Tao Li; Yanjie Song

cs.AINov 3, 2026

Large Language Model-Assisted Adaptive Heuristic Search for Agile Earth Observation Satellite Scheduling

Authors: He Wang, Yeye Liu, Ruijie Li, Xuemei Jiang, Junyu Wu, Lining Liu Zhuoqi Man, Tao Li, Yanjie Song

Organizations: College of Intelligent Science and Engineering, Harbin Engineering University, Harbin 150001, China · School of Electrical and Control Engineering, North University of China, Taiyuan 030051, China · School of Electrical and Electronic Engineering, Shandong University of Technology, Zibo, 255000, China · School of Unmanned Aerial Vehicle Industry, Chengdu Aeronautic Polytechnic University, Chengdu 610042, China ·

Abstract

Agile Earth observation satellites must coordinate imaging tasks within limited observation windows, while accounting for attitude maneuvering and onboard energy constraints. Competition among observation opportunities makes scheduling performance sensitive to the heuristic rules used for task selection. This letter proposes a large language model (LLM)-assisted adaptive heuristic search method for agile Earth observation satellite scheduling. The LLM generates an initial heuristic rule according to the characteristics of the scheduling instance. During the subsequent ant colony search, search performance and scheduling information are fed back to the LLM to refine the heuristic rule. The proposed modifications are evaluated through paired solution constructions before being incorporated into the search process. A unified scheduling procedure determines task selection, satellite assignment, and observation timing under the operational constraints. Experiments on six scheduling scenarios involving 50--200 imaging tasks demonstrate that the proposed method achieves 4.3%--9.6% higher mean observation benefit than the strongest baseline.

Figures & tables

Panel of Figure 1 — Fig. 1: Feasible observation insertion and alternative opportunities.

Fig. 2: Framework of the proposed LLM-assisted adaptive heuristic search for AEOSs scheduling.

Scenario	ACO	ABC	GA	PSO	WOA	LLMACO	Gain vs. best (%)
S1	$225.70\pm 4.82$	$228.10\pm 3.51$	$227.30\pm 5.06$	$233.55\pm 3.17$	$221.05\pm 5.24$	$\mathbf{243.65\pm 0.81}$	$+4.32^{+}$
S2	$181.90\pm 3.52$	$191.05\pm 2.82$	$184.40\pm 7.14$	$191.85\pm 4.39$	$175.75\pm 6.34$	$\mathbf{202.50\pm 0.51}$	$+5.55^{+}$
S3	$489.80\pm 6.01$	$486.00\pm 5.89$	$489.60\pm 9.91$	$503.75\pm 8.75$	$476.10\pm 7.85$	$\mathbf{528.20\pm 1.58}$	$+4.85^{+}$
S4	$416.35\pm 5.59$	$416.05\pm 5.83$	$419.05\pm 8.99$	$431.10\pm 9.87$	$397.80\pm 8.91$	$\mathbf{470.35\pm 0.75}$	$+9.10^{+}$
S5	$844.05\pm 7.27$	$817.90\pm 9.16$	$835.60\pm 11.84$	$866.05\pm 12.19$	$812.15\pm 12.61$	$\mathbf{920.55\pm 2.26}$	$+6.29^{+}$
S6	$690.60\pm 13.29$	$675.50\pm 7.98$	$683.65\pm 10.93$	$716.20\pm 9.88$	$660.25\pm 10.08$	$\mathbf{785.20\pm 2.21}$	$+9.63^{+}$

TABLE I: Comparison of observation benefit after 100 iterations (mean

\pm

standard deviation over 20 independent runs). The last column reports the improvement over the strongest baseline. Best values are highlighted in bold.

Fig. 3: Convergence comparison of different algorithms under scenario S6.

Fig. 4: Runtime distributions of different algorithms under scenarios S1, S4, and S6.

Language Model-Guided Search Heuristics

See all 23 papers

Optimizing Earth Observation Satellite Schedules under Unknown Operational Constraints: An Active Constraint Acquisition Approach
Apr 14, 2026Mohamed-Bachir BelaidConstrained Optimization Learning to Optimize
NILU, Climate and environmental research institute, Kjeller, Norway
EOS-Bench: A Comprehensive Benchmark for Earth Observation Satellite Scheduling
Apr 28, 2026Qian Yin, Jiaxing Li, Jiaqi Cheng +23Benchmark Construction Benchmark Design
School of Traﬃc and Transportation Engineering, Central South University, Changsha 410083, China · School of Engineering and Materials Science, Queen Mary University of London, London E1 4NS, UK · College of Automation, Central South University, Changsha, 410083, China +20
Reinforcement Learning-Guided Evolutionary Policy Optimization for Preference-Adjustable Heterogeneous Agile Earth Observation Satellite Scheduling
Aug 25, 2026He Wang, Junyu Wu, Hui Li +3RL for Combinatorial Optimization Multi-Objective Evolutionary Optimization
College of Intelligent Science and Engineering, Harbin Engineering University, Harbin 150001, China · School of Information Science and Technology, Dalian Maritime University, Dalian 116026, China · Silesian University of Technology (SUT), Department of Measurement and Control Systems, Gliwice, Akademicka 2, 44-100 Poland +3

Apr 14, 2026J/K move · Enter open · S save

Optimizing Earth Observation Satellite Schedules under Unknown Operational Constraints: An Active Constraint Acquisition Approach

Mohamed-Bachir Belaid

NILU, Climate and environmental research institute, Kjeller, Norway

Constrained Optimization Learning to Optimize Remote Sensing

Open paper AskPDF arXiv

Earth Observation (EO) satellite scheduling (deciding which imaging tasks to perform and when) is a well-studied combinatorial optimization problem. Existing methods typically assume that the operational constraint model is fully specified in advance. In practice, however, constraints governing separation between observations, power budgets, and thermal limits are often embedded in engineering artefacts or high-fidelity simulators rather than in explicit mathematical models. We study EO scheduling under unknown constraints: the objective is known, but feasibility must be learned interactively from a binary oracle. Working with a simplified model restricted to pairwise separation and global capacity constraints, we introduce Conservative Constraint Acquisition~(CCA), a domain-specific procedure designed to identify justified constraints efficiently in practice while limiting unnecessary tightening of the learned model. Embedded in the Learn&Optimize framework, CCA supports an interactive search process that alternates optimization under a learned constraint model with targeted oracle queries. On synthetic instances with up to 50tasks and dense constraint networks, L&O improves over a no-knowledge greedy baseline and uses far fewer main oracle queries than a two-phase acquire-then-solve baseline (FAO). For $n\leq 30$ , the average gap drops from 65--68% (Priority Greedy) to 17.7--35.8% using L&O. At $n{=}50$ , where the CP-SAT reference is the best feasible solution found in 120s, L&O improves on FAO on average (17.9% vs.\ 20.3%) while using 21.3 main queries instead of 100 and about

5\times

less execution time.

Abstract

Language Model-Guided Search Heuristics

Explore similar work

Optimizing Earth Observation Satellite Schedules under Unknown Operational Constraints: An Active Constraint Acquisition Approach

EOS-Bench: A Comprehensive Benchmark for Earth Observation Satellite Scheduling

Reinforcement Learning-Guided Evolutionary Policy Optimization for Preference-Adjustable Heterogeneous Agile Earth Observation Satellite Scheduling