Kaiyan Chang, Yonghao Shi, Chenglong Wang, Hang Zhou, Chi Hu, Xiaoqian Liu, Yingfeng Luo, Yuan Ge, Tong Xiao, JingBo Zhu. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

We present a new approach to address the challenges associated with maintaining the functionality of triboelectric vibration sensors in smart plywood during woodworking operations involving nails and screws. The current state-of-the-art sensor design employs non-overlapping electrodes, which unfortunately leads to significant compromises in terms of signal strength and clarity, particularly in real-world scenarios that involve electromagnetic (EM) interference. To overcome these limitations, we propose a method that enables the woodworker to manually isolate short-circuited electrodes. This method facilitates the creation of sensors using overlapping electrodes, while also incorporating EM shielding, thereby resulting in a substantial improvement in the sensor’s robustness when detecting user activities. To validate the effectiveness of our proposed approach, we conducted a series of experiments, which not only shed light on the drawbacks of non-overlapping electrode designs but also demonstrated the significant improvements achieved through our method.

Tagnoo is a computational plywood augmented with RFID tags, aimed at empowering woodworkers to effortlessly create room-scale smart environments. Unlike existing solutions, Tagnoo does not necessitate technical expertise or disrupt established woodworking routines. This battery-free and cost-effective solution seamlessly integrates computation capabilities into plywood, while preserving its original appearance and functionality. In this paper, we explore various parameters that can influence Tagnoo’s sensing performance and woodworking compatibility through a series of experiments. Additionally, we demonstrate the construction of a small office environment, comprising a desk, chair, shelf, and floor, all crafted by an experienced woodworker using conventional tools such as a table saw and screws while adhering to established construction workflows. Our evaluation confirms that the smart environment can accurately recognize 18 daily objects and user activities, such as a user sitting on the floor or a glass lunchbox placed on the desk, with over 90% accuracy.