Research Area

We research the full-stack neuromorphic system by following the workflow from Neuromorphic Device modeling and Software Algorithm optimization to Hardware System implementation. This cycle ensures that device-level characteristics are perfectly reflected in the system's real-time operation. 

Our work features the integration of Digital Logic (FPGAs) with Analog Memcapacitor Arrays. By developing the entire stack—including the Driver Layer, Middleware, and Application Layer—we provide  Neuromoprhic AI hardware accelerators for next-generation computing.

We research a sensor-fused intelligent systems by integrating multiple sensing modalities. Input data streams from a diverse array of sensors—including Ultrasonic, RADAR, LiDAR, and Vision—are collected to provide a multi-layered view of the environment. 

The "Sensor Fusion + AI Hardware" processing core then merges and interprets this heterogeneous data to enable real-time decision-making. This core is designed for high-performance and low-latency operation, potentially utilizing our custom AI accelerator IPs and scalable hardware platforms like NPUs and FPGAs by bridging open frameworks like ONNX and leading solutions from industry leaders like AMD and Nvidia. 

We specialize in the design and implementation of high-speed digital logic IPs optimized for high-performance signal processing and precision measurement. Our research involves developing various signal processing IP such as hardware-efficient accelerators for Multiply-Accumulation (MAC) and Fast Fourier Transform (FFT) algorithms, and also focus on high-resolution Time-to-Digital Converters (TDCs) and ranging calculators, which are critical for advanced sensing applications like LiDAR and RADAR. These custom IPs are verified through rigorous FPGA floorplanning and implementation to guarantee Reliable performance in demanding and mission-critical environments.