Industrial edge AI platform integrates multimodal acceleration

Edge AI platforms are increasingly being designed to process multimodal data locally to support real-time decision making in industrial automation and intelligent infrastructure. In this context, Kontron introduced the KBox A-151 EAI, an industrial edge AI computer developed in collaboration with SiMa.ai for high-performance AI inference in edge environments.

Edge AI architecture for multimodal industrial workloads
The KBox A-151 EAI is designed as an industrial edge computing platform for multimodal control, autonomous systems and AI inference tasks. The system combines 13th Gen Intel^® Core™ or Intel Atom^® processors with the SiMa.ai MLSoC™ Modalix AI accelerator to deliver more than 50 TOPS (tera operations per second) of AI processing performance.

The dual-processor architecture separates conventional application processing from AI inference tasks, allowing operational logic and edge AI workloads to run independently. This architecture is intended to support system stability in industrial environments where deterministic processing behaviour is required.

Application areas include industrial automation, energy infrastructure, medical technology and intelligent transportation systems, where edge processing can reduce latency and dependence on cloud connectivity within distributed digital infrastructure.

Hardware design for deployment in industrial environments
The system uses a fanless, passively cooled enclosure designed to maintain thermal performance without active cooling components. This approach reduces maintenance requirements and avoids performance limitations associated with thermal throttling.

The platform supports real-time video analytics and AI workloads including generative AI models, large language models (LLMs), large multimodal models (LMMs) and convolutional neural networks (CNNs). It also supports end-to-end machine vision pipelines and multi-sensor data integration.

Integration with the Palette™ software development kit and Edgematic™ development environment allows deployment of multimodal machine learning models using standard machine learning frameworks. This approach supports model deployment without major hardware redesign, which can reduce engineering effort during development cycles.

Enabling physical AI in edge environments
The platform targets physical AI applications where AI systems must interpret sensor data and interact with physical processes in real time. This includes combining vision, audio and other sensor inputs into operational decision workflows.

The KBox A-151 EAI includes configurable I/O interfaces, scalable memory configurations and industrial certification support to enable deployment in environments requiring long product lifecycles and predictable system behaviour.

By integrating the SiMa.ai MLSoC™ accelerator into an industrial edge platform, the system is positioned to support AI inference performance alongside the reliability requirements typically associated with industrial computing platforms.

www.kontron.com

Edited by industrial journalist, Aishwarya Mambet — AI-powered.