Fanless Mini METIS - IP51
• Fanless Edge AI Computer • Model: FM-METIS-51
A compact, fanless edge AI system engineered for industrial and outdoor deployments. Purpose-built to run multi-stream computer vision inference at the edge, it delivers reliable, low-maintenance AI performance where space, power efficiency and continuous operation matter most.
FEATURES
A fanless platform for controlled edge AI.
The FM-METIS-51 is a compact, fanless edge AI platform that combines the Axelera Metis AI Processing Unit with an ARM-based host processor to perform computer vision and AI inference directly at the edge. Designed for real-time object detection, classification, segmentation and video analytics, it processes live camera streams locally to reduce bandwidth consumption, minimise latency and keep sensitive data on-site. Typical deployment scenarios include retail, traffic and mobility, logistics, security, smart infrastructure and Industry 4.0 applications.
AI acceleration - Axelera Metis® AIPU
Powered by the Axelera Metis® AI Processing Unit, a quad-core AI accelerator purpose-built for edge AI inference. It enables real-time object detection, classification, tracking, segmentation and multi-stream video analytics directly where data is generated, reducing bandwidth consumption, minimizing latency and keeping sensitive data on-site.
Host processor
An ARM-based Rockchip RK3588 host processor works alongside the Axelera Metis® AIPU to run the operating system, application software and the pre- and post-processing stages of the AI inference pipeline, providing a balanced platform for high-performance edge AI workloads.
AI software - Voyager® SD
An ARM-based Rockchip RK3588 host processor works alongside the Axelera Metis® AIPU to run the operating system, application software and the pre- and post-processing stages of the AI inference pipeline, providing a balanced platform for high-performance edge AI workloads.
Fanless, passively cooled design
A fully fanless, passively cooled architecture eliminates moving parts, reducing power consumption and maintenance requirements. Without fans to draw dust and debris into the enclosure, the platform is well suited to reliable long-term operation in industrial environments while offering improved resistance to shock and vibration.
Connectivity
Dual Gigabit Ethernet ports are provided as standard, with optional Wi-Fi or 4G/LTE connectivity for networked and remotely managed edge AI deployments.
IP51-rated enclosure
The IP51-rated enclosure helps protect internal components from dust ingress and light dripping water, supporting reliable operation in industrial environments. Its sealed design reduces the accumulation of dust inside the system, helping minimise maintenance and extend long-term reliability.
Chassis & sustainability
The in-house designed and precision-machined aluminium chassis is engineered for efficient passive cooling, strength and long-term durability. Manufactured using repurposed and recycled aluminium wherever possible, with production waste recycled as part of the manufacturing process, it supports a more sustainable approach to edge AI hardware.
IDEAL DEPLOYMENTS
Where the FM-METIS-51 is ideal.
Engineered for fixed edge AI deployments in industrial and indoor environments, supporting continuous multi-stream computer vision with reliable, low-maintenance operation.
Industrial manufacturing
Supports automated visual inspection, defect detection, predictive maintenance and workplace safety monitoring, enabling real-time quality control and operational visibility on the production line.
Retail
Enables self-checkout object recognition, real-time shelf and inventory monitoring, and loss prevention through automated shrinkage detection and store analytics.
Security & surveillance
Supports multi-camera incident detection, restricted-area monitoring and PPE compliance verification, enabling real-time security and safety decisions directly at the point of capture.
Logistics
Supports movement tracking, throughput monitoring and workplace safety across warehouses, distribution centres and logistics yards, providing real-time operational visibility and process optimization.
THE CURRENT GENERATION
FM-METIS-51 Fanless Edge AI Computer
A summary of the platform's core capability. The full technical specification, I/O layout, and options are available on the datasheet.
Specifications | |
Processor | Rockchip RK3588 (ARM) - 8-core, up to 2.4 GHz |
AI accelerator | Axelera Metis® AIPU |
AI Performance | 214 TOPS |
Memory | 16 GB LPDDR4X |
Storage | 64 GB eMMC (option: up to 8 TB SATA SSD) |
Networking | 2 × Gigabit LAN (optional Wi-Fi / 4G-LTE connectivity via internal M.2) |
Operating temperature | 0 °C to 60 °C |
Dimensions / Weight | 51 × 177 × 208 mm · 2.4 kg |
For illustrative purpose
Rear I/O: DC-In 12V, 2x RJ45 for LAN, 4x USB 3.2, 1x HDMI, 1x USB-C, Micro SD Slot
ENGINEERING & SUSTAINABILITY
Engineered in-house for performance and sustainability
The chassis is designed and precision-machined in-house, not sourced as an off-the-shelf enclosure. Its geometry is optimised for efficient passive cooling, structural strength and long-term reliability, with a design proven across extended operating temperatures.
The same engineering approach supports sustainability. The chassis is manufactured using repurposed and recycled aluminium wherever possible, while aluminium waste generated during production is recovered and recycled, reducing material waste without compromising performance.
- In-house designed and milled aluminium chassis
- Optimized for efficient passive cooling
- Proven reliability across extended operating temperatures
- Repurposed and recycled aluminium
- Production waste recycled
FREQUENTLY ASKED QUESTIONS.
The FM-METIS-51 is built on the Axelera Metis® AIPU, a quad-core AI Processing Unit using Axelera’s Digital In-Memory Computing (D-IMC) architecture, purpose-built for computer-vision inference at the edge.
The Metis AIPU is programmed with Axelera’s Voyager® SDK, which supports common frameworks PyTorch, ONNX and TensorFlow (via ONNX) alongside a library of pre-trained models. Model preparation follows Axelera’s pipeline-based workflow.
Yes. A single Metis AIPU can run multiple concurrent video streams and several parallel neural networks, which suits multi-camera analytics. Actual stream counts depend on the models and resolution used.
Yes. Inference runs locally, so data is processed where it is created reducing bandwidth use, keeping data on-site, and removing dependence on a continuous cloud connection.
Computer-vision tasks, object detection, classification, tracking, segmentation and video analytics.
Controlled options, operating-system deployment, software preload and branding are available and do not alter the certified platform baseline. Requirements that need mechanical, thermal, electronic or certification change route to Tranquil Studio.
Tell us what you need your AI to do.
Share your use case, deployment environment and operational constraints, and we'll recommend the edge AI compute platform that's right for your application whether powered by Axelera Metis or NVIDIA Jetson. From standard platform selection to controlled configuration, every deployment is governed from day one.