Multispectral AI at the Hong Kong Electronics Fair

The HKTDC Hong Kong Electronics Fair (Autumn Edition) 2026, hosted at the Hong Kong Convention and Exhibition Centre, has become a key venue for showcasing advancements in artificial intelligence and robotics. Among the notable participants, Shenzhen HQVT Technology Co., Ltd. (stock code: 1392.HK) has introduced its multispectral AI fire early warning portfolio. This technology is designed to address safety challenges in critical infrastructure, including data centres, power distribution systems, and new energy facilities.

By presenting a workflow that connects abnormal signal detection with on-site verification and response, the company aims to demonstrate how physical AI can be applied to industrial safety. The exhibition, which runs from 13 to 16 October, features more than 3,200 exhibitors from 15 countries and regions, highlighting a global interest in automated and intelligent safety solutions.

Addressing Critical Infrastructure Risks

Modern industrial facilities such as data centres, power distribution systems, and new energy facilities share several operational characteristics that make fire prevention challenging. These environments feature densely installed equipment, complex electrical systems, and a high risk of severe consequences if a fire or equipment failure occurs. The impact of such incidents often extends beyond physical damage to include business interruption, asset losses, and safety risks to personnel.

In these settings, identifying risks before they escalate is highly valuable. Many electrical faults produce detectable physical signals long before visible smoke or flames appear. For example, a loose electrical connection may cause abnormal discharge or localised heating. Because these weak signals are difficult to detect through conventional visible-light monitoring alone, operators require advanced sensing technologies to identify early signs of risk.

The Optics-Sensor-Imaging-Computing Architecture

To address these challenges, HQVT has developed a proprietary “Optics-Sensor-Imaging-Computing” technical architecture. This system combines multiple sensing methods with artificial intelligence to analyze environmental risks:

  • Ultraviolet Sensing: This method is used to detect electrical discharge and early-stage flames, which are often invisible to the human eye.
  • Infrared Data: This component monitors temperature distribution and identifies abnormal heat buildup in electrical components.
  • Visible-Light Imaging: This provides the necessary scene context for operators to understand the physical environment.

By analyzing these signals together, the multispectral AI system helps identify early risk indicators that conventional visible-light systems may miss. During the exhibition, the company demonstrated how this technology converts invisible physical signals into actionable risk indicators, allowing operators to locate and investigate potential hazards.

A Tiered Hardware Ecosystem for Diverse Deployments

The portfolio exhibited in Hong Kong demonstrates how on-site sensing, local analysis, and personnel response can be integrated. To support different installation conditions and deployment requirements, the company has developed several specialized terminals:

  • Industrial Terminal: Designed for monitoring physical safety in standard industrial environments.
  • Explosion-Proof Analyser: Built to address hazardous or specialized installation conditions where explosion risks exist.
  • Mini Card-Style Terminal: Intended for smaller sites and lightweight deployments where space is limited.
  • Ceiling-Mounted Privacy-Preserving Terminal: Utilizes ultraviolet and infrared sensing to monitor risks without capturing visible-light images, making it suitable for privacy-sensitive areas.

These terminals are designed to identify electrical discharge, abnormal heat, and small flames across various operating environments.

Integrating Edge Computing and Robotics

Beyond static sensing terminals, the workflow incorporates edge computing and mobile robotics to enable a complete response loop. The exhibition features the Zhiyuan Edge Hub, which plays a role at the computing level by processing data locally. This local analysis is crucial for physical AI, helping machines interpret real-world conditions and generate rapid alerts.

Additionally, the portfolio includes a quadruped inspection robot equipped with the company's multispectral AI module. This robotic integration demonstrates how automated systems can perform on-site verification and patrol high-risk areas, bridging the gap between initial signal detection and physical response. By combining stationary sensors, edge computing hubs, and mobile robotic units, the system provides a multi-layered approach to industrial safety and risk management.

As buyers from the Asia-Pacific region, Europe, and South America engage in discussions at the exhibition, the deployment of multispectral AI highlights a growing industry focus on proactive, sensor-driven safety systems for high-density electrical and energy infrastructure.