Overview
Geoscience research requires complex analysis and domain expertise, with remote sensing (RS) observations as a key foundation. However, existing RS agents built on general-purpose LLMs remain largely domain-agnostic, leading to brittle and error-prone workflows. Moreover, these failures are seldom consolidated into reusable experiences for subsequent analyses. To address this issue, we introduce RSMeM, a knowledge-enhanced memory evolution mechanism that bootstraps RS agents with pre-distilled domain knowledge and iteratively integrates online experience for robust multi-step tool execution.
Mechanism Components
RSMeM is composed of two main components:
- Hierarchical Knowledge Grounding: This component performs taxonomy-aware retrieval over a hierarchical domain corpus to guide planning and tool selection;
- Failure-Aware Experience Refinement: This component distills failure-annotated tool-use traces into reusable constraints for next-round tool execution.
By iteratively employing these two processes, RS agents can evolve to absorb task-level domain knowledge and effectively translate it into instance-level execution experience.
Experimental Results
Extensive experiments on EarthBench demonstrate that RSMeM consistently improves tool-use performance and end-to-end accuracy across a diverse set of LLM backbones. Notably, RSMeM achieves a 6% accuracy improvement on DeepSeek-V3.2 with less than 1% additional experience tokens, showcasing the strong knowledge density of our distilled experience. Our code is available at GitHub.
Blogger's Review: The RSMeM mechanism significantly enhances the performance of remote sensing agents through the combination of knowledge grounding and experience refinement, highlighting the importance and potential of domain knowledge. This approach not only improves model adaptability but also offers new insights for future remote sensing analysis.