RESEARCH

Architectures Our lab explores computer architecture innovations for artificial intelligence. We focus on memory-centric computing to minimize data movement and overcome the memory wall. We also design neural processing units with high energy efficiency and scalability. By combining these approaches, we enable optimized AI training and inference from edge to datacenter.

Systems Embodied AI applications require intensive computation for perception, reasoning, and interaction in real-world environments. In contrast, embedded systems operate under strict resource constraints in terms of power, memory, and processing capability. Bridging this gap calls for specialized computing techniques that balance performance efficiency with hardware limitations.

Computing Our lab optimizes artificial intelligence and robotic frameworks for efficient execution. We focus on reducing memory footprint and improving runtime performance in resource-constrained systems. Through hardware–software co-design, we accelerate critical modules for real-time operation. These efforts enable lightweight, scalable, and high-performance solutions for AI and robotics.