Marvel Digital AI Limited
Metax Technologies Co., Limited
Simbury Limited
Edge device AI computing is one of the most promising area of innovation in IC design today. The focus of edge device AI is caused by the rapid growth of applications that require real-time data processing, such as smart wearables, robots, healthcare, and self-driving cars. On-device data processing has a high requirement on computation performance, energy efficiency, and chip area. For some advanced applications, edge devices are required to support complex functions. For example, the emerging smart in-car central control system may contain the functions of face recognition, gesture detection, road/object detection, and fatigue detection. This project aims to design an ASIC-based AI core to accelerate the multiple functions while maintaining low power and low cost. To achieve this goal, we optimize the design at various levels including hardware design, compilation, and quantization. We will also conduct rigorous RTL and system-level verification to ensure reliability. The IP core will be taped-out for mass production and will be adopted in smart in-car central control systems. It can also be applied to other smart home and smart city applications. The developed solution will benefit the SMEs in deploying edge AI related applications.