Edge Computing Systems for Urban Video Analytics

Advantech is introducing a portfolio of edge artificial intelligence platforms designed to process real-time urban infrastructure data at the sensor level. These localized computing solutions target intelligent video analytics, physical safety monitoring, and autonomous mobility systems within municipal environments.

The hardware will be exhibited at the Smart City Expo World Congress (SCEWC) 2026 in Barcelona, Spain, from November 3 to 5. The computing platforms are engineered to convert raw sensor feeds into actionable metrics by executing computer vision algorithms locally on the device, thereby reducing latency and minimizing continuous bandwidth reliance on centralized cloud networks.

Neural Processing Hardware for Urban Visual Data
The company's Video Analytics Platform (VSS) integrates NVIDIA Jetson Thor architecture to execute visual data analysis directly at the edge for public safety applications. To support distributed artificial intelligence deployments, the Next-Gen Smart CityMind platform utilizes the AFE-A702 hardware hub to function as a vision-language agentic controller. This hub interfaces directly with the EdgeSight Sentinel system, which operates on the AFE-R750 edge node powered by NVIDIA Jetson AGX Orin. This establishes a structural hierarchy where localized field nodes stream continuous visual monitoring data and edge-inferred metadata to a central directing hub for complex scenario assessment.

Infrastructure Integration and LiDAR Traffic Analytics
For public transportation applications, the technology integrates with existing closed-circuit television (CCTV) cameras to avoid the cost of full hardware replacement. Partnering with software developer Technis, the deployment uses the TREK-50N vehicle computing system paired with NVIDIA Jetson Orin NX to execute real-time object classification and event detection directly on buses and trains. In the domain of traffic flow analysis, the MIC-733 computing platform processes 3D LiDAR spatial data generated by Ouster Gemini and BlueCity sensors. This configuration utilizes CUDA and TensorRT acceleration to calculate measurable traffic density and velocity analytics without relying on traditional optical camera tracking.

Autonomous Robotics and Digital Infrastructure Scalability
Expanding the digital infrastructure across public spaces, the MIC-770 and MIC-785 GPU platforms, featuring RTX PRO 4000 Blackwell architecture, support high-density video analytics applications developed by Macnica and Icetana AI for commercial retail areas. Additionally, robust edge computing enables localized spatial navigation for autonomous robotics. By integrating the MIC-732 industrial system with an NVIDIA Jetson AGX Orin 64GB module, self-driving utility robots can rapidly process multi-modal sensor inputs to navigate securely in dynamic, unpredictable real-world environments without requiring persistent external server connectivity.

Additional Context
This section details technical specifications and competitive benchmarking not included in the original news release.

Within the industrial edge computing market, AI inference platforms are evaluated based on their thermal operating range, ingress protection ratings, and total neural processing throughput measured in Tera Operations Per Second (TOPS). Competitors such as Kontron and Axiomtek offer comparable ruggedized edge gateways utilizing NVIDIA Jetson and Intel processor architectures for machine vision tasks. Differentiation among these digital infrastructure solutions is objectively determined by the efficiency of hardware acceleration for specific deep learning frameworks, the availability of high-bandwidth I/O interfaces capable of handling multiple continuous video streams, and adherence to established shock and vibration standards required for demanding transit and robotic deployments.

Edited by Aishwarya Mambet, Induportals Editor, with AI assistance.

www.advantech.com