Email / GitHub / Google Scholar / X

Yaozhong (York) Kang is an M.E. student at the Institute of Science Tokyo, advised by Prof. Kazuhiro Nakadai. He is interested in motion planning methods that generalize across sensing modalities and embodiments, and in building the fast simulators they are trained in.

Previously, York earned his bachelor degree in Automation from Southern University of Science and Technology (SUSTech). During his bachelor studies, he was working on navigation systems for all kinds of robots.

Publications

What Do Neural Networks Learn for TDOA Estimation? A Cross-Architecture Probing Study Y. Kang, J. Wang, R. Shi, T. Ashizawa, B. Yen, K. Nakadai
Interspeech 2026, arXiv, Code Cross-power computation emerges in neural networks trained for TDOA estimation.

BCNav: Bearing-Conditioned Depth Policies for Sound Source Navigation Y. Kang, J. Wang, T. Ashizawa, B. Yen, K. Nakadai
IEEE RO-MAN 2026, IEEE, arXiv Replacing acoustic simulation with sound source localization lets a behavior-cloned motion-planning depth policy navigate to sound on a real robot.

Observability-Aware Active Calibration of Multisensor Extrinsics for Ground Robots via Online Trajectory Optimization J. Wang†, Y. Kang†, L. Fu, K. Nakadai, H. Kong
IEEE Sensors Journal 2025, IEEE, arXiv, Code Plans robot trajectories that maximize the observability of microphone-array, LiDAR and wheel-encoder extrinsics, calibrating them online.

Collaborative Fall Detection and Response Using Wi-Fi Sensing and Mobile Companion Robot Y. Chen†, Y. Kang†, Z. Zhao, Y. Hong, L. Meng, M. Q.-H. Meng
IEEE ROBIO 2024, IEEE, arXiv Detects falls out of line of sight from Wi-Fi channel state information and sends a mobile dual-arm robot to respond, opening a door with OMPL.

A Constant-Time Acoustic Cue for Robot Learning Y. Kang, K. Nakadai
第44回日本ロボット学会学術講演会(RSJ 2026)

† denotes co-first authors.

Services

Reviewer: RO-MAN, ICME, ICRA, T-RO, CASE, SII, Interspeech