Papers
34
Total Citations
620
H-Index
14
About
Noriaki Hirose is a robotics and machine learning researcher whose work spans vision-based robot navigation, traversability estimation, social robotics, and deep learning-driven control systems. He has made significant contributions to the development of intelligent autonomous robots that can navigate complex, dynamic environments without relying on traditional mapping and localization pipelines. Hirose's most influential work includes "Deep Visual MPC-Policy Learning for Navigation" (2019, 88 citations), which demonstrated how robots could follow image-defined trajectories in a manner reminiscent of human landmark-based navigation. His semi-supervised deep learning framework GONet (2019, 71 citations) advanced traversability estimation using fisheye cameras and Generative Adversarial Networks, enabling safer robot deployment in unstructured environments. He also contributed to GNM (2023, 71 citations), a generalizable navigation model trained across diverse robot platforms, pushing the boundaries of data-efficient learning for robotics. Beyond navigation, Hirose has explored social robot behavior through SoPhie and SACSoN, addressing human-compliant path prediction and scalable social navigation. His earlier work on personal assistive robots and friction modeling via LSTM neural networks reflects a strong foundation in physical control systems. Collectively, his research demonstrates a sustained commitment to bridging perception, learning, and real-world robot deployment.
Research Focus
Key Achievements
Top Papers
- 1Deep Visual MPC-Policy Learning for Navigation88 citations · 2019
- 2GNM: A General Navigation Model to Drive Any Robot71 citations · 2023
- 3
- 4
- 5Modeling of rolling friction by recurrent neural network using LSTM42 citations · 2017
- 6
- 7SACSoN: Scalable Autonomous Control for Social Navigation31 citations · 2023
- 8
- 9
- 10MPC policy learning using DNN for human following control without collision16 citations · 2018