摘要：用于航空观测的边缘平台必须在内存有限、算力有限且连接间歇的条件下解读飞机图像。对于标准的仅 RGB 识别模型和通用视觉语言模型而言，这一场景颇具挑战，尤其是在缺少经过校准的热成像和 LiDAR 飞机数据时。我们提出 TriCLE，这是一套面向应用、适用于边缘约束条件下飞机分类群组划分的三模态视觉语言系统。TriCLE 可从单张 RGB 飞机图像生成保持结构的 FLIR 风格热成像视图及伪 LiDAR 深度投影，随后在紧凑型 Qwen3-VL 主干模型中，将这些对齐视图与任务指令融合。该模型与一套由专家定义的飞机分类体系对齐，该体系基于推进方式、机身家族、尺寸、设计年代和构型，因此其输出反映的是与工程相关的相似性，而不仅仅是表面外观。我们评估了监督微调、旋转保持型 SFT，以及三种策略对齐方法：GRPO、GSPO 和 DAPO。序列级 GSPO 取得了最强的验证性能，在有效飞机输出上达到 88.33% 的验证准确率和 0.91 的加权 F1。在留出的飞机测试划分上，GSPO 达到 78.00% 的准确率和 0.793 的加权 F1，同时保持 94.00% 的输出格式可解析性。经过 4 比特量化和注意力内存优化后，完成对齐的 4B 模型可满足 8GB 部署目标，并以每个三模态样本 1.48 秒的速度完成处理。这些结果表明，TriCLE 可作为一种实用原型，用于可解释且适合边缘部署的飞机分组；同时也强调，仍需在真实且对齐的热成像与 LiDAR 传感器数据流上进一步验证。

Abstract:Edge platforms used for aerial observation must interpret aircraft imagery under limited memory, limited compute, and intermittent connectivity. This setting is difficult for standard RGB-only recognition models and general-purpose vision-language models, especially when calibrated thermal and LiDAR aircraft data are unavailable. We present TriCLE, an application-oriented tri-modal vision-language system for aircraft taxonomic grouping under edge constraints. From a single RGB aircraft image, TriCLE generates a structure-preserving FLIR-style thermal view and a pseudo-LiDAR depth projection, then fuses the aligned views with task instructions in a compact Qwen3-VL backbone. The model is aligned to an expert aircraft taxonomy based on propulsion, airframe family, size, design era, and configuration, so its outputs reflect engineering-relevant similarity rather than only surface appearance. We evaluate supervised fine-tuning, rotation-preserving SFT, and three policy-alignment strategies: GRPO, GSPO, and DAPO. Sequence-level GSPO gives the strongest validation performance, reaching 88.33% validation accuracy and 0.91 weighted F1 on valid aircraft outputs. On a held-out aircraft test partition, GSPO achieves 78.00% accuracy and 0.793 weighted F1 while preserving 94.00% parseable output formatting. After 4-bit quantization and attention-memory optimization, the aligned 4B model fits an 8GB deployment target and processes each tri-modal triplet in 1.48 seconds. These results support TriCLE as a practical prototype for interpretable, edge-feasible aircraft grouping, while emphasizing the need for further validation on real aligned thermal and LiDAR sensor streams.