Kioxia America, Inc. announced a collaboration with Kioxia Corporation, Tsubakimoto Chain Co., and EAGLYS Inc. to develop an AI-powered image recognition system aimed at automating product identification in logistics workflows. The initiative leverages KIOXIA AiSAQ software and Memory-Centric AI technology, designed to improve efficiency, scalability, and adaptability in increasingly complex supply chains. The technology will be showcased at the 2025 International Robot Exhibition in Tokyo from December 3–6.
The project responds to growing challenges in logistics, driven by the surge in e-commerce transactions and expanding product diversity. Labor shortages and increasing operational demands have made automation essential, but traditional image recognition systems struggle to keep pace. Standard AI models require frequent retraining whenever new or seasonal products are introduced, increasing both processing time and energy costs.
KIOXIA AiSAQ and Memory-Centric AI address these limitations by enabling high-capacity storage of new product data, including images, labels, and features, without needing to retrain base AI models. The technology also employs data indexing and SSD storage to accelerate retrieval in AI systems using Retrieval Augmented Generation (RAG), allowing rapid classification even as product catalogs expand.
The collaboration demonstrates the potential for enhanced operational efficiency in logistics. At the exhibition, visitors will see a conveyor-based system that captures images of products in real time and classifies them by referencing stored feature data. The demonstration highlights how logistics operators can manage a continually changing and wide-ranging inventory with greater speed and accuracy.
KIOXIA’s approach underscores a broader trend toward memory-centric AI solutions, which reduce retraining needs, optimize power usage, and enable scalable AI adoption across industries. By integrating advanced storage solutions with AI-driven analytics, the technology provides logistics operators with a pathway to automated, cost-efficient, and highly adaptable workflows.
