Date & Time: 

   2026 /10 / 16  (Fri) 14:20 - 16:20

 

Location: 

   Delta Building R216, NTHU

 

Speaker: 

  溫志煜 教授

  國立中興大學 電機工程學系

 

Topic: 

  When Audio AI Meets Low-Cost Sensing: High-Privacy Two-Target Indoor Tracking via Conv-TasNet and PIR Analog Signatures

 

Abstract: 

  Passive infrared (PIR) sensors are highly ideal for indoor positioning because they are cost-effective, consume low power, and inherently protect user privacy. However, existing PIR systems primarily target single-occupant scenarios; separating mixed analog signals from co-occurring individuals remains a critical bottleneck. This talk introduces a groundbreaking framework designed to overcome this multi-target challenge. By adapting Conv-TasNet—a deep learning network originally designed for audio source separation—the system successfully isolates overlapping PIR signals from two distinct targets. To resolve post-separation identity ambiguity, a joint Mel-frequency cepstral coefficients (MFCC) and XGBoost classification model is introduced to correctly map the signals. Experimental evaluation in a standard 3.3 × 3.3 m² environment demonstrates that the proposed MFCC-XGBoost model achieves an impressive 73% tracking accuracy and a mere 0.39-meter path error. This breakthrough reduces localization error by 42% compared to current state-of-the-art multi-target systems, all while maintaining a real-time inference latency of just 0.07 seconds. These findings prove that advanced algorithms can unlock high-precision, real-time multi-target tracking on cost-effective hardware, offering a highly practical and privacy-preserving solution for next-generation smart environments.

 

Autobiography: 

  Chih-Yu Wen (Senior Member, IEEE) received the B.S.E.E. and first M.S.E.E. degrees in electrical engineering from National Cheng Kung University, Tainan, Taiwan, in 1995 and 1997, respectively, and the second M.S.E.E. and Ph.D. degrees in electrical engineering from the University of Wisconsin–Madison, Madison, WI, USA, in 2002 and 2005, respectively. He joined the Department of Electrical Engineering, National Chung Hsing University, Taichung, Taiwan, in 2006, where he is currently a Lifetime Distinguished Professor. His current research interests include wireless communications, biomedical signal processing for health monitoring, and distributed networked sensing and control. He was a recipient of the National Innovation Awards–Institute for Biotechnology and Medicine Industry, in 2016, 2018, 2019, and 2025, for contributions to remote pulmonary rehabilitation and smart medical devices. He and his co-authors were also recipients of the Best Paper and Oral Presentation Awards, such as the 2020 Taiwan Telecommunications Annual Symposium Best Paper Award, the 2021 IEEE ECBIOS Best Paper Award, and the 2025 IEEE ICCE-TW Best Oral Presentation Award at “Smart Systems and Its Applications” Session. From 2018 to 2025, he was an Associate Editor of IET Signal Processing and International Journal of Fuzzy Systems.