According to the 2025 Intelligent Manufacturing Logistics Efficiency Survey, the logistics links of domestic large-scale manufacturing enterprises account for an average of 37% of the total production time, and 62% of efficiency losses stem from decision-making delays caused by the lack of edge-side computing power.
When the production rhythm is compressed to the minute level, the transmission latency, bandwidth bottleneck and data security risks of cloud centralized computing power have become the core structural obstacles restricting the production capacity leap of manufacturing enterprises. Building a new intelligent logistics system with "distributed edge-side computing power sinking" has become a rigid demand for industrial upgrading.
Adhering to the core principle of "scenario adapts to computing power, and computing power drives decision-making", this solution relies on edge computing terminals of different levels to build a three-level AI computing power network covering storage nodes, transfer nodes and regional dispatching centers, completely realizing local closed-loop processing of full-process logistics data.
At the terminal nodes of warehouse operations, the USR-EG218 embedded ARM industrial PC is deployed as the front-end perception computing carrier. Equipped with a high-performance AI processing unit, it can deliver a real-time inference computing power of up to 46 TOPS. A single device can connect 32 channels of high-definition vision acquisition terminals in parallel, and realize unified scheduling of data streams and video streams relying on the built-in WukongEdge engine.
In the scenario of precision component warehousing, the device can run three algorithms (OCR barcode recognition, appearance defect detection, and intelligent storage location allocation) in parallel locally, without returning video data to the cloud. The verification latency for a single product is controlled within 200 ms, which boosts efficiency by 90% compared with the traditional cloud processing mode, while avoiding the security risk of core production material data leakage.
At the mobile nodes of factory transfer, the USR-EG928A AI edge computing box is used as the vehicle-mounted computing terminal. Adopting an 8-core 64-bit industrial-grade processor, it provides a stable 6 TOPS computing power output, adapts to the complex vibration and wide-temperature environment on vehicles, and supports native docking with existing MES and WMS systems of manufacturing enterprises via Ubuntu.
When deployed on unmanned transfer vehicles and industrial forklifts, it can run three edge algorithms in real time: personnel intrusion warning, cargo posture monitoring, and route deviation identification. The response latency for abnormal events is less than 50 ms, which eliminates the transfer loss of valuable production materials from the source. Actual measurement data from deployed enterprises shows that the incidence of factory logistics safety incidents drops by 92% after the deployment of this node.
At the dispatching hub node of factory logistics, the USR-EG528 rack-mounted AI vision industrial PC is adopted as the regional computing core. With an expandable high-performance computing module, it supports a maximum comprehensive computing power of 46 TOPS, and is equipped with multiple industrial-grade Gigabit Ethernet ports and RS485/232 serial ports as standard, enabling unified access and data collaboration of dozens of types of logistics terminals across the entire factory.
Relying on its powerful local computing power, it can dynamically run the global multi-AGV path optimization algorithm, and independently generate the globally optimal transfer scheme by combining multi-dimensional variables such as real-time production order priority, equipment load, and factory road conditions. This can reduce the empty driving rate of factory logistics vehicles by 17% and increase the overall logistics turnover efficiency by 35%.
Different from traditional intelligent logistics solutions, this edge AI computing power logistics system delivers irreplaceable core values in three dimensions:
Efficiency Value: The full-link local computing closed loop eliminates cloud transmission latency, cutting the overall logistics operating cost by 42%.
Security Value: All core production logistics data remains in the enterprise intranet throughout the process, fully meeting the data compliance requirements of manufacturing enterprises.
Scalability Value: The device is pre-installed with an iterable AI algorithm library of over 200 types, which can continuously extend the functional life cycle as the enterprise's production scenarios upgrade, providing a sustainable computing power base for the long-term digital transformation of intelligent manufacturing enterprises.
