Embedded Edge AI platforms for vision, voice and sensor application

Customer: International technology OEM developing intelligent and connected industrial products

Project: Development of two embedded Edge AI solutions for local AI processing in connected products

Product Description

Syntronic has developed two embedded Edge AI solutions designed to bring AI processing directly into products and devices.

The solutions are based on the ArtInChip D133CBS and Rockchip RV1126B platforms, providing different levels of processing capability for applications ranging from smart cameras and AIoT devices to industrial interfaces, voice assistants and intelligent connected products.

By processing data locally, the platforms can reduce response times and dependence on cloud infrastructure. They can also enable products to continue operating with limited or no network connectivity while keeping sensitive data within the device.

The solutions are designed around practical Edge AI applications rather than requiring the computing resources of large generative AI models. Depending on the platform and application, they can support functions such as image classification, object and human detection, voice wake-up, keyword recognition, gesture recognition and intelligent sensor analysis.

Scope

The development included hardware integration and platform development around two different processor architectures, with the aim of supporting different Edge AI requirements.

The first solution is based on the ArtInChip D133CBS and combines lightweight AI processing with graphics, multimedia and real-time control capabilities. The platform is suited for applications such as smart displays, smart-home controllers, voice-interaction devices, AI cameras, industrial HMIs and lightweight robotics.

The second solution is based on the Rockchip RV1126B and provides greater processing capacity for more demanding AI and computer vision applications. The platform combines a quad-core Cortex-A53 processor with dedicated AI acceleration and image-processing capabilities.

The Rockchip-based solution can also support lightweight language and multimodal models running locally. Its software ecosystem provides a development path toward more powerful platforms such as the RK3576 and RK3588 if additional processing capacity is required in later product generations.

The development was also supported by Syntronic's broader product development capabilities, including schematic design, PCB layout, thermal simulation, environmental testing, EMC/EMI testing and support for CE, FCC and CCC certification.

Results

The result is two flexible Edge AI platforms that provide different levels of local processing for AI-enabled products.

The ArtInChip-based solution provides a compact foundation for products combining lightweight AI with graphics, multimedia and real-time control. The Rockchip-based solution provides additional processing capacity for AI vision, AIoT and applications requiring more demanding local models.

Both platforms allow data to be collected, processed and analysed directly on the device. This can support faster response times, reduce the need for continuous cloud connectivity and provide greater control over how data is handled.

The platforms also provide a starting point for companies that already have AI algorithms but need hardware and embedded engineering expertise to bring them into a production product.

By combining Edge AI platforms with hardware development, embedded systems expertise and product industrialisation capabilities, Syntronic can support the development process from early prototype through New Product Introduction (NPI) and production.