Architecture of an AI on Device Board for Local, Automated Training and Inference

This paper presents the conceptualization and implementation of an AI on Device HAT Sensor-board for the Raspberry Pi aimed for model inference and data collection, coupled with a shared storage unit for data exchange. With increasing complexity of artificial intelligence applications and the resulting cost of energy, data, and hardware requirements, assessments of resource- and energy-efficiency are often neglected. We attempt to address these areas with the help of TinyML based solutions. We demonstrate the implementation of the AI on Device Board without external transfer requirements, present the hardware and software concept, and implement a use case. Additionally, we show how these methodologies can help to achieve a higher grade of energy and resource efficiency while maintaining the desired performance. Furthermore, the transfer of existing methodologies to an AI on Device board is demonstrated in an exemplary use case to show the adaptability of the proposed hardware.