Event Classification of Accelerometer Data for Industrial Package Monitoring with Embedded Deep Learning

2025-06-05

AI Summary

To address long-term (multi-year) structural health monitoring of reusable industrial packaging, this paper proposes a lightweight embedded deep learning framework tailored for ultra-low-power IoT devices. To tackle the challenges of small-sample, multi-class, and class-imbalanced accelerometer time-series data, we integrate SMOTE and ADASYN for synthetic time-series augmentation. A compact 1D-CNN architecture is designed and optimized via structured pruning and post-training quantization, achieving 4× model compression (75% reduction in memory footprint). The system enables real-time edge inference for critical handling events—forklift lifting and truck transportation—with binary classification accuracies of 94.54% and 95.83%, respectively, while consuming only 316 mW during inference and exhibiting significantly reduced wake-up latency. Our key contribution is the first end-to-end embedded AI pipeline specifically engineered for the multi-year lifecycle of industrial packaging—uniquely balancing high accuracy, robustness to operational variability, and ultra-low power consumption.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Package monitoring is an important topic in industrial applications, with significant implications for operational efficiency and ecological sustainability. In this study, we propose an approach that employs an embedded system, placed on reusable packages, to detect their state (on a Forklift, in a Truck, or in an undetermined location). We aim to design a system with a lifespan of several years, corresponding to the lifespan of reusable packages. Our analysis demonstrates that maximizing device lifespan requires minimizing wake time. We propose a pipeline that includes data processing, training, and evaluation of the deep learning model designed for imbalanced, multiclass time series data collected from an embedded sensor. The method uses a one-dimensional Convolutional Neural Network architecture to classify accelerometer data from the IoT device. Before training, two data augmentation techniques are tested to solve the imbalance problem of the dataset: the Synthetic Minority Oversampling TEchnique and the ADAptive SYNthetic sampling approach. After training, compression techniques are implemented to have a small model size. On the considered twoclass problem, the methodology yields a precision of 94.54% for the first class and 95.83% for the second class, while compression techniques reduce the model size by a factor of four. The trained model is deployed on the IoT device, where it operates with a power consumption of 316 mW during inference.

Problem

• Classify package states using embedded deep learning on accelerometer data
• Maximize device lifespan by minimizing wake time for industrial monitoring
• Address data imbalance with augmentation and compress model for IoT deployment

Innovation

• Embedded deep learning for package state detection
• Data augmentation to address dataset imbalance
• Model compression for reduced size and power