Chan-Wei Hu

Papers

Total Citations

H-Index

About

Chan-Wei Hu is a researcher specializing in computer vision, autonomous navigation, and hardware-accelerated machine learning, with particular expertise in applying deep learning to real-world robotic and embedded systems challenges. Hu's most recognized contribution is the development of the Omnidirectional Convolutional Neural Network (O-CNN), a novel architecture designed to tackle the demanding problem of visual place recognition under severe camera pose variation. By leveraging omnidirectional cameras, this work addresses scenarios where only limited place exemplars are available — a common and critical constraint in practical navigation systems. The paper has garnered over 74 citations, reflecting meaningful influence within the robotics and computer vision communities. Beyond perception, Hu has extended research interests into the intersection of reinforcement learning and hardware efficiency, as demonstrated by the TD3lite project, which explores FPGA acceleration of deep reinforcement learning algorithms through structural and representational optimizations — a contribution aimed at making computationally intensive RL techniques viable for resource-constrained platforms. Collectively, Hu's work bridges the gap between advanced machine learning methodologies and their practical deployment in autonomous systems, making meaningful contributions to both algorithmic innovation and efficient hardware implementation.