KIOXIA Unveils AI-Driven Image Recognition System for Next-Generation Logistics Automation, Powered by AiSAQ™ and Memory-Centric AI Technologies
Kioxia Corporation today announced a significant technological breakthrough with the development of an advanced AI-based automatic image recognition system designed to improve efficiency and accuracy in modern logistics environments. Developed in collaboration with Tsubakimoto Chain Co. (Tsubakimoto Chain) and EAGLYS Inc. (EAGLYS), the jointly engineered solution automatically identifies a wide range of products and packages as they move through logistics workflows. This innovative system is expected to enable logistics operators to accelerate automation, reduce labor demands, and maintain high service quality in an increasingly complex and fast-changing market landscape.
The newly developed system is built around KIOXIA AiSAQ™ and Memory-Centric AI technologies—two core components intended to address the scalability challenges faced by organizations attempting to adopt AI across expanding product catalogs and variable supply chain environments. The technology will be demonstrated publicly for the first time at the 2025 International Robot Exhibition, providing industry stakeholders with a hands-on preview of its capabilities and operational advantages.
Transforming Logistics Through Intelligent Automation
The logistics industry is undergoing profound transformation driven by the rapid acceleration of e-commerce, global trade expansion, and evolving consumer expectations for faster, cheaper, and more predictable delivery. The volume and variety of goods moving through global logistics networks continue to increase, placing strain on infrastructure, workforce resources, and operational processes.
Simultaneously, labor shortages have emerged as a critical challenge across the logistics sector. Warehouses, fulfillment centers, and distribution hubs face increasing difficulty hiring and retaining skilled workers while maintaining productivity targets and service quality levels. As a result, organizations are turning to AI-powered automation to improve efficiency, reduce errors, and enable workforce optimization.
However, traditional image recognition systems have struggled to keep pace with real-world logistics demands. Most deep learning–based image classification solutions require extensive retraining and parameter adjustment whenever new product types enter the supply chain or existing products undergo packaging changes. In industries such as retail, fashion, grocery, electronics, and seasonal merchandise, product variations may change weekly or even daily. This constant need for retraining introduces:
- Higher operational costs
- Increased energy consumption
- Implementation delays
- Frequent system downtime
- Reliability risks in fast-moving environments
The result is an industry-wide need for flexible, scalable AI solutions that can quickly adapt to dynamic product environments without the heavy burden of retraining core models.0
A Breakthrough Approach: KIOXIA AiSAQ™ and Memory-Centric AI
KIOXIA’s new solution directly addresses these limitations through its intelligent Memory-Centric AI architecture, which allows large-scale product information—including images, labels, and product feature sets—to be stored directly in high-capacity storage rather than embedding all data into a trained model.
Instead of rebuilding neural network parameters every time logistics content changes, the system enables new product information to be:
- Collected and stored quickly in memory,
- Indexed for rapid similarity-based search, and
- Referenced by the recognition engine in real time without retraining the base model.
In other words, the system continuously evolves without performance disruption—allowing operators to manage endless product variety and unpredictable inventory conditions.
As stored product data increases, traditional AI systems typically face longer retrieval times and significantly higher memory consumption. The KIOXIA approach solves this challenge by indexing stored data and migrating the indexed data to high-performance SSD storage, enabling faster access and scalable operational efficiency even as product volumes grow.
This results in reduced power consumption, faster processing speeds, and lower total cost of ownership—crucial advantages in large-scale logistics operations.
Real-World Demonstration at the 2025 International Robot Exhibition
KIOXIA, Tsubakimoto Chain, and EAGLYS will showcase the jointly developed image recognition technology at the 2025 International Robot Exhibition, held December 3–6 at Tokyo Big Sight (Tsubakimoto Chain Booth E6-23). As one of the world’s most influential robotics and automation technology events, the exhibition provides an ideal venue for demonstrating leading-edge innovations shaping next-generation industrial automation.
During live on-site demonstrations, attendees will observe the product recognition system operating within a realistic logistics line environment. Products transported along a conveyor system will be scanned, identified, and classified using camera capture and stored feature references. The system will display how it rapidly recognizes items even when new products are introduced or when surface appearances vary due to packaging changes or seasonal variations.
The demonstration will illustrate:
- Rapid image capture and recognition
- Flexible adaptation to diverse product types
- Elimination of model retraining for new items
- Reduced processing time even with large datasets
- Practical scalability for real-world logistics environments
In environments where product mixes may change weekly or even daily, this system enables significantly more agile operations, supporting efficient warehouse automation, inventory accuracy, order fulfillment, and supply chain responsiveness.
A New Era of Intelligent Logistics Infrastructure
The global logistics industry is undergoing a foundational shift from labor-intensive processes to more intelligent, automated, and digitally connected systems. AI technologies—particularly in recognition, decision support, and predictive analytics—are expected to become central infrastructure elements that determine the competitiveness and resilience of logistics networks.
The development of this new AI-powered image recognition system represents a milestone in that transition. By effectively eliminating the need for constant AI model retraining, the solution helps organizations:
- Reduce operational costs
- Improve processing speed and classification accuracy
- Respond quickly to product variability and demand fluctuations
- Accelerate technology deployment timelines
- Improve workplace safety by automating repetitive tasks
- Enhance productivity even with smaller labor forces
The collaboration between Kioxia, Tsubakimoto Chain, and EAGLYS brings together strengths in storage technology, industrial automation, and secure AI-based data processing. KIOXIA contributes advanced memory and storage solutions; Tsubakimoto Chain brings decades of expertise in material handling and conveyor automation; and EAGLYS provides AI innovation and secure data infrastructure.
Together, the three organizations aim to deliver scalable, real-world AI solutions capable of supporting the rapidly evolving needs of global commerce.
Source Link:https://www.businesswire.com/news
