Modern workplaces and mission-critical environments increasingly rely on cameras to understand what is happening around them. Yet collecting video is only the first step. To support faster decisions, intelligent cameras need to process visual information close to where it is captured, while meeting demanding requirements for security, integration and scalability.
Edge vision AI brings intelligence directly to cameras and connected systems, reducing the need to send every video stream to a central location. This approach can support real-time awareness across vehicles, robots, facilities and other operational environments. The challenge is building as highlighted in ST Engineering's advancements in enhancing safety with smarter technology.
Traditional camera deployments often begin with hardware selection, followed by the addition of AI capabilities. This can create integration challenges when cameras, computing systems and AI models have not been designed to work together from the start. For organisations operating large or complex fleets, the result can be higher system complexity and greater demands on computing and network resources.
An AI-native approach changes this model by considering vision, processing and deployment requirements together. It allows intelligent vision systems to be tailored to specific applications while supporting scalable deployment. This is especially relevant where systems must operate within vehicles, unmanned platforms, critical sites or facilities with limited compute and bandwidth.
A scalable vision system needs to balance image quality, processing requirements, physical constraints and cost. Using different camera configurations for different operational needs can help organisations avoid applying the same hardware design to every environment. It also allows the vision solution to evolve as applications become more demanding.
ST Engineering's PAVE platform was developed around this principle. It combines specialty cameras developed in-house with commercial off-the-shelf options, allowing solutions to be tailored to different applications. Its low-distortion, wide field-of-view camera technology is designed to reduce the number of cameras needed while lowering video-processing demands, helping AI models run efficiently at the edge.
PAVE is available in three configurations designed for different operating requirements:
| Configuration | Resolution | Example Applications |
|---|---|---|
| PAVE Lite | 5MP | FPV drones, vehicles and critical site monitoring |
| PAVE Pro | 12MP | UGVs and AI surveillance |
| PAVE Ultra | >20MP | Demanding, high-performance applications |
The platform began as a ground-up innovation in 2023 and has since been integrated into several ST Engineering products. The company is also engaging local and overseas customers as edge vision AI moves toward wider operational adoption. The different configurations provide a practical way to align camera capability with the needs of manned, unmanned and high-performance systems.
PAVE began as an innovative camera supporting land systems. It has grown to become a capable edge vision AI solution suitable for many more mission-critical applications. Edge vision AI is starting to be adopted at scale, and I believe we are well-positioned to ride this wave of growth!
Chin Kai Lun, Head of Emerging Business, ST Engineering
Processing visual data at the edge can be valuable when systems need to respond quickly to changing conditions. Instead of relying on continuous transmission of video to a remote processing environment, AI models can analyse information closer to the camera and application. This can help support real-time performance where bandwidth, compute capacity or connectivity may be constrained.
For mobile platforms, this capability can be particularly important. Vehicles and robots may need to interpret their surroundings while operating in dynamic environments, while facilities may require continuous monitoring without creating unnecessary data-processing demands, enabling more efficient data and AI workflows at the edge. By reducing video-processing requirements through a wide field of view, intelligent camera systems can help make edge AI more practical for these use cases.
Edge vision AI is not limited to one type of workplace or platform. Its value can extend across operational environments where visual information needs to be interpreted quickly and reliably.
This flexibility reflects a wider shift in how organisations approach computer vision. Rather than treating cameras as passive devices that simply capture footage, intelligent systems can make visual information more useful at the point of operation. The right architecture can help organisations deploy AI across different platforms without creating a separate vision stack for every use case.
Real-world deployment requires more than strong AI models. Cybersecurity, system integration and trusted supply chains are also important considerations, particularly for systems operating in sensitive or mission-critical environments. Future development of the PAVE platform includes embedded cybersecurity components, advanced night vision and specialised cameras designed for amphibious, desert and arctic conditions.
Keeping deployed systems current is another important consideration. Lightweight AI models and field-updatable tools can help organisations adapt capabilities after deployment rather than relying only on major hardware replacements. Collaboration with local and international startups, SMEs and internal AI teams also supports continued development across specialised vision and AI capabilities.
The growth of intelligent cameras highlights an important lesson for organisations exploring AI: successful deployment depends on the complete system, not just the AI model. Camera design, computing requirements, cybersecurity, integration and operating conditions all influence whether a vision solution can perform reliably in the field.
ST Engineering's work on PAVE reflects this system-level approach, bringing camera technology and edge AI together for scalable applications. From manned vehicles and unmanned platforms to facilities and mission-critical environments, intelligent vision can help turn visual data into timely operational insight.
Edge vision AI enables organisations to process visual data directly where it is captured, supporting real-time awareness while keeping compute and bandwidth demands manageable. By taking an AI-native approach, matching camera capabilities, operational resilience, and cybersecurity to specific mission needs, ST Engineering’s configurable PAVE platform provides a practical, scalable foundation for vehicles, unmanned systems, and critical infrastructure alike. To learn how PAVE can transform your edge vision operations, contact the ST Engineering team today.
By subscribing to the mailing list, you confirm that you have read and agree with the Terms of Use and Personal Data Policy.