TinyML让立方体卫星在低功耗下实现高效图像分类。
TinyML Enhances CubeSat Mission Capabilities
- 用剪枝+量化+硬件适配压缩模型，适配微型处理器。
- 内存降89.55%，闪存降70.09%，精度损失仅0.4~8.6个百分点。
- 适合资源受限的航天任务，尤其关注能效与实时性。
地球观测任务通常将原始或初步处理的影像传回地面站进行计算密集型分析，但这一模式难以适用于立方体卫星，因其对机载嵌入式处理器、能源和通信带宽有严格限制。本文提出一种基于TinyML的卷积神经网络（ConvNets）模型优化与部署流程，支持在立方体卫星级约束下实现高精度、低功耗且硬件感知的推理。该流程整合结构化迭代剪枝、训练后INT8量化及硬件感知算子映射，将模型压缩并适配STMicroelectronics的STM32N6微控制器（集成Arm Cortex-M55核心与神经处理单元NPU），以模拟真实立方体卫星计算架构。在EuroSAT、RS_C11、MEDIC三个地球观测基准数据集上，对SqueezeNet、MobileNetV3、EfficientNet、MCUNetV1四种模型进行评估。结果显示，优化后模型平均减少RAM使用89.55%、Flash存储70.09%，显著降低下行链路带宽需求，同时保持任务可接受的准确率（相比浮点基线下降0.4至8.6个百分点）。单次推理能耗为0.68~6.45 mJ，延迟为3.22~30.38 ms，完全满足立方体卫星所需的严苛能效与实时性要求。
原文摘要 · Abstract (English)
Earth observation (EO) missions traditionally rely on transmitting raw or minimally processed imagery from satellites to ground stations for computationally intensive analysis. This paradigm is infeasible for CubeSat systems due to stringent constraints on the onboard embedded processors, energy availability, and communication bandwidth. To overcome these limitations, the paper presents a TinyML-based Convolutional Neural Networks (ConvNets) model optimization and deployment pipeline for onboard image classification, enabling accurate, energy-efficient, and hardware-aware inference under CubeSat-class constraints. Our pipeline integrates structured iterative pruning, post-training INT8 quantization, and hardware-aware operator mapping to compress models and align them with the heterogeneous compute architecture of the STM32N6 microcontroller from STMicroelectronics. This Microcontroller Unit (MCU) integrates a novel Arm Cortex-M55 core and a Neural-ART Neural Processing Unit (NPU), providing a realistic proxy for CubeSat onboard computers. The paper evaluates the proposed approach on three EO benchmark datasets (i.e., EuroSAT, RS_C11, MEDIC) and four models (i.e., SqueezeNet, MobileNetV3, EfficientNet, MCUNetV1). We demonstrate an average reduction in RAM usage of 89.55% and Flash memory of 70.09% for the optimized models, significantly decreasing downlink bandwidth requirements while maintaining task-acceptable accuracy (with a drop ranging from 0.4 to 8.6 percentage points compared to the Float32 baseline). The energy consumption per inference ranges from 0.68 mJ to 6.45 mJ, with latency spanning from 3.22 ms to 30.38 ms. These results fully satisfy the stringent energy budgets and real-time constraints required for efficient onboard EO processing.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。