Iván Hernández Dalas: Advantech introduces edge AI systems for a range of robot embodiments
Advantech said its latest systems bring the power of NVIDIA’s Jetson Thor platform to real-world applications. | Source: Advantech
Advantech, a developer of edge computing technology, introduced a new lineup of application-focused Edge AI systems powered by NVIDIA Jetson Thor modules. The company is targeting real-world robotics applications through hardware-software integrated systems for robotics, medical AI, and data AI.
Each offering features application-specific hardware platforms, pre-integrated with JetPack 7.0, remote management tools, and vertical software suites such as Robotic Suite and GenAI Studio. Built on a container-based architecture, these systems offer greater flexibility and faster development cycles, Advantech said.
Advantech said the NVIDIA Jetson Thor series set a new benchmark for edge AI, delivering up to 2070 FP4 TFLOPS of AI performance, along with significant improvements in CPU performance and energy efficiency.
In addition to providing NVIDIA Jetson Thor boards, systems, and software design-in services for vertical solutions, Advantech collaborates closely with ecosystem partners on key technologies such as sensor and camera integration, as well as thermal design. This holistic approach empowers developers to build and deploy edge AI applications faster, more easily, and more efficiently, the company claimed.
Taipei, Taiwan-based Advantech develops IoT intelligent systems and embedded platforms, with a focus on edge computing and edge AI. The company targets five key markets: edge intelligence systems, manufacturing, energy and utilities, healthcare, and city services and retail.
Advantech offers robotic controllers for humanoids, AMRs, AVs, and surgical robots
Advantech purpose-built the ASR-A702 and AFE-A702 robotic controllers for humanoids, AMRs, and unmanned vehicles. They deliver real-time AI reasoning and inference with GPU-accelerated SLAM, supporting multi-camera GMSL, 2D/3D sensors, and IMUs.
With Robotic Suite for plug-and-play development, plus Isaac ROS/Sim and Holoscan for real-time perception and ultra-low latency data flows, they enable rapid integration and deployment.
Key features include hardware time sync, ESD protection, anti-vibration design, and OTA upgrades— ensuring stable, safe, and high-performance computing across smart logistics, service robotics, and mission-critical unmanned applications.
By leveraging NVIDIA Jetson Thor with advanced SDKs such as Holoscan and MONAI, Advantech empowers next-generation Medical AI board AIMB-294 and system EPC-T5294. These platforms accelerate real-time sensor processing, image analysis & streaming AI pipeline, pre-trained model, and 3D imaging optimization, and surgical robotics focus with low latency and high precision for operating rooms, clinical workflows, and intelligent diagnostic tools.
Advantech aims to bring LLMs to the edge
AIR-075 delivers powerful computing with 4× 10GbE and GMSL interfaces to satisfy Data AI demands in traffic and factory applications. Combined with NVIDIA AI, NVIDIA Metropolis, NVIDIA Triton, NVIDIA Cosmos Reason, and Advantech Edge AI SDK & DeviceOn, it enables sensor fusion, multi-model inference, a visual AI agent, and centralized management for real-time, predictive edge intelligence.
Advantech Container Catalog (ACC) delivers a cluster of ready-to-develop edge AI applications, including end-to-end computer vision and Edge LLM environments optimized for AI agent integration on NVIDIA Jetson platforms. It also offers domain-specific solutions from ecosystem partners—such as robotics perception, surgical imaging, healthcare, and smart city sensing—enabling rapid deployment across industrial and vertical markets.
Fully compatible with WEDA (WISE-Edge Developer Architecture), its containerized architecture enables scalable edge AI expansion, from single-node setups to distributed edge networks.
The post Advantech introduces edge AI systems for a range of robot embodiments appeared first on The Robot Report.
View Source
