Po‐Kai Hsu

National Taipei University of Technology

Papers

1

Total Citations

8

H-Index

1

About

Po-Kai Hsu is a researcher advancing the field of intelligent robotics and computer vision, with a primary focus on real-time topological localization and deep learning integration. His most cited work, "Real-time topological localization using structured-view ConvNet with expectation rules and training renewal" (2020), has garnered 8 citations and introduces a novel framework that combines structured-view convolutional neural networks with expectation-based rules and adaptive training renewal. This contribution addresses critical challenges in robotic navigation by enabling robust, real-time place recognition in dynamic environments, bridging the gap between theoretical deep learning models and practical deployment. Hsu’s approach emphasizes efficiency and adaptability, making it particularly valuable for autonomous systems requiring continuous learning and minimal computational overhead. Beyond this flagship paper, his research explores the intersection of spatial reasoning, neural network optimization, and sensor fusion, with implications for service robotics, autonomous vehicles, and smart infrastructure. While his citation count reflects an emerging career, the methodological innovation in his work—particularly the integration of expectation rules for model refinement—positions him as a promising voice in applied AI. Hsu’s contributions offer a pragmatic pathway toward more resilient and self-improving robotic systems, appealing to students and researchers interested in bridging theory with real-world autonomy.

Research Focus

Key Achievements

1
H-Index
1
Papers
8
Total Citations
8
Avg Citations/Paper
🏆 Most Cited Paper
Real-time topological localization using structured-view ConvNet with expectation rules and training renewal
8 citations · 2020
📈 Most Prolific Year: 2020 (1 Papers)
🤝 Key Collaborators: 3
🏛 Institutions: National Taipei University of Technology

Top Papers

  1. 1

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 11 days ago