Seongjin Wang

Seongjin Wang

Auto-generated profile, based on publicly available information

Seongjin Wang

Yonsei University, Seoul, Republic of Korea

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 February 2024
Verified
Conference Paper
Open access
CKCherry Kim
Cherry Kim
SWSeongjin Wang
Seongjin Wang
DKDongjun Kim
Dongjun Kim
Shijia PanShijia Pan
Shijia Pan
JHJun Han
Jun Han
Visually impaired individuals face challenges in navigating unfamiliar indoor environments, which typically feature complex layouts and various furniture. Although several solutions have been proposed to assist them, there are still limitations. For instance, WiFi-based localization methods tend to lose accuracy due to interference issues, requiring additional deployment costs to achieve higher accuracy [2]. Moreover, smart canes equipped with proximity sensors often fail to provide precise location information and involve additional hardware, increasing the weight of the cane [3]. To address these limitations, we propose ImpacTile, a tile-level localization system, which identifies a user location by analyzing the sound produced when users drop their cane on a floor tile without requiring extra infrastructure. Figure 1(a) depicts a usage scenario of ImpacTile. The core idea of ImpacTile is that distinctive acoustic signatures are generated across each tile. This is possible due to the distinct physical properties under each tile (e.g., variations in concrete cracks and applied adhesive shapes) [1]. The distinct reverberation at each location further enhances these variations. We present the design of ImpacTile in Figure 1(b). However, designing ImpacTile poses several challenges. First, the impact sound produced across different tiles often exhibits similarities, which can reduce the accuracy of our system. We address this with a Pre-processing and Feature Extraction module that focuses on identifying structural differences. Second, pinpointing the exact location in large buildings with numerous tiles can be challenging. To mitigate this, we employ an Adjacency-based Enhancement module, where the user drops the cane on two adjacent tiles. This leverages the adjacency information of the tiles to enhance accuracy. To verify the feasibility of ImpacTile, we conduct experiments across three buildings: Building 1 (325 tiles), Building 2 (331 tiles), and Building 3 (500 tiles). We recruit two participants to drop the cane on the center of each tile with five trials per tile. Each participant uses a different cane and smartphone model while recording at 48 kHz sampling rate. In the Bootstrapping Phase, one participant gathers acoustic data from every tile on a single floor of each building. Subsequently, in the Deployment Phase, the other participant collects data from 100 random tiles in each building. Figure 1(c) demonstrates the performance of ImpacTile in three buildings, both before and after the Adjacency-based Enhancement module. The results indicate that while ImpacTile already achieves high accuracy before the enhancement, its accuracy further improves after applying the enhancement. Subsequently, we anticipate expanding the applications of ImpacTile to assist visually impaired users. We envision extending ImpacTile into an indoor turn-by-turn navigation system, not only on tiled floors but also on non-tiled floors.


4 November 2024
Verified
Conference Paper
Open access
DSDongjin Seo
Dongjin Seo
JYJonghyuk Yun
Jonghyuk Yun
SWSeongjin Wang
Seongjin Wang
JSJaewoo Son
Jaewoo Son
JHJun Han
Jun Han
Continuous monitoring of patients in psychiatric seclusion rooms is essential yet challenging, particularly with staff shortages that can delay responses to sudden changes in patient conditions. To this end, we propose PsiMo, a remote patient state monitoring system using mmWave Frequency Modulated Continuous Wave (FMCW) radar. Unlike existing vision-based or wearable systems, PsiMo captures patient movements without compromising privacy or risking potential self-harm. Our system continuously monitors patient's state of motion to alert medical staff in the event of abnormal conditions, such as agitation. Our preliminary evaluation shows PsiMo achieves 97.0% accuracy in patient state classification, demonstrating its potential for effective, non-contact monitoring.