Aetina Showcases Physical AI, Vision AI, and Agentic AI Solutions at COMPUTEX 2026
May 27, 2026
News
During COMPUTEX 2026, Aetina Corporation will demonstrate its latest innovations in edge AI solutions in Hall 1, Booth K0106. The booth will highlight Aetina’s portfolio including SuperEdge, MegaEdge, DeviceEdge and CoreEdge along with real-time interactive demonstrations co-developed with strategic partners and ecosystem innovators. The exhibition will show how edge AI is progressing from single inference models into integrated applications with contextual intelligence across Physical AI, Vision AI, and Agentic AI.
Physical AI:
The main booth attraction is Aetina’s AI-powered robotic arm system designed for the edge and developed with NVIDIA and Franka Robotics. It is integrated with technology from Axis, Innodisk, PrefactorTech, and Stereolabs and will demonstrate how edge AI can be implemented for next-generation robotics, smart factories, autonomous systems, and human-machine interaction.
The solution integrates Aetina’s DeviceEdge AIB-AT78/68, powered by the NVIDIA Jetson Thor module, with Aetina’s SuperEdge AEX-2UA1 GPU server accelerated by NVIDIA RTX PRO 6000 Blackwell Server Edition GPU. An on-site Private LLM and NVIDIA ACE allows a digital avatar to engage visitors through real-time voice interaction, while a Stereolabs ZED camera supports object recognition and 3D perception. In its entirety, the workflow runs at the edge without cloud dependency, demonstrating a full-stack physical AI architecture where voice, vision, reasoning and motion control operate concurrently.
On display will be a Humanoid Surround Vision Solution built on the NVIDIA Jetson Thor platform that integrates Stereolabs ZED X stereo cameras and ZED X One cameras through the high-bandwidth, low-latency GMSL2 interface. This provides real-time depth sensing, peripheral awareness and human-like navigation capabilities for industrial robots and embodied AI applications.
Aetina, Digital Media Professionals Inc. (DMP), Cambrian Robotics, and Universal Robots will collaborate to exhibit an Automated Wiring and Cable Assembly Solution powered by the Aetina MegaEdge AIP-KQ67 edge AI inference platform. The solution utilizes Cambrian’s AI vision perception to guide a dual-arm (UR) robotic system in high-precision wire harness assembly. It allows robots to recognize, guide, and insert flexible cables and connectors in datacenter server assembly, electronics, and automotive production lines.
Vision AI
Aetina and ISP will demonstrate a lightweight VLM application powered by the palm-sized DeviceEdge AIE-CN33/43-A1 Mini system and leveraging NVIDIA Jetson Orin NX 16GB module with integrated DRAM-on-module. The Jetson delivers a scalable platform supporting enhanced runtime inference with minimal memory footprint for leading open generative AI models. When combined with ISP’s gLupe Watcher technology, the solution reduces the memory footprint needed by VLM models. It enables compact edge devices to accomplish near server-grade visual recognition and is ideal for construction site safety monitoring, smart inspection, and anomaly detection.
Also highlighted will be the DeviceEdge AIE-KR1B-B1, an Arm-based edge AI system powered by the Rockchip processor. The solutions modularized M.2 architecture supports diverse AI accelerator cards, including Axelera AI, DEEPX, EdgeCortix, MemryX and REALTEK ideal for multimodal generative AI and AI vision applications.
Agentic AI
Aetina will focus on the flagship MegaEdge AIP-FR68S workstation series utilizing the Qualcomm DragonwingTM AI On-Prem Appliance Solution supporting compute power and stability involved in large language model deployment and advanced AI vision inference.
Aetina, Qualcomm, and MoBagel will showcase GenieAcceler, an enterprise AI solution combining on-premises deployment, Agentic RAG, and source citation to allow secure knowledge search, professional document generation, and autonomous workflow execution.
Collaborating with Data Systems Co., Ltd., Aetina will showcase the AI Entry Agent combining multimodal AI agents and OCR technology to identify document images and automatically extract and enter data, converting them into structured data integrated into ERP systems. It also features adaptive learning and error correction capabilities delivering an end-to-end intelligent workflow for automated order generation.
For more information, visit aetina.com/.
Chad Cox is the Production Editor at Embedded Computing Design. His responsibilities are centered around content creation, writing and editing, and article research and development. Chad covers industry news and events and is known to interact with various industrial leaders via on-premise visits and online interviews. He is responsible for the digital footprint and dissemination of news via social media posts, advertising creation and the production of newsletters including the Embedded Computing Design’s Daily.
He is well versed in many facets of industrial computing including Edge AI, IoT, Processing, Security, Open Source, and more.
Chad graduated from the University of Cincinnati with a B.A. in Cultural and Analytical Literature and holds a master’s in education.
