We proposed a modular plug-and-play sensor system for urban air pollution monitoring that adopting the dedicated Modular Sensor System (MSS) architecture and Universal Sensor Interface (USI), and modular design in a sensor node. Such system has expandable sensor modules with plug-and-play feature and supports multiple Wireless Sensor Networks (WSNs, e.g., Bluetooth, ZigBee, GPRS, Wi-Fi. Depending on the intended application, the user can configure the sensor node simply by inserting the sensor modules and wireless data link. The system will then handle the data acquisition and transmission automatically.

A compact sensor node with a Bluetooth Low Energy (BLE) data link and eight plug-and-play sensor modules, namely CO, SO2, O3, NO2, CO2, temperature-humidity-pressure (THP), and radiation (NR1, NR2), have been implemented. Dedicated mobile application, web application and back-end server have also been implemented for data visualization, management, and analysis.

We have also collaborated with oversea researchers from Korea Institute of Science and Technology Information (KISTI) and developed a system integrating the PM2.5 data provided by governmental stations and vehicular sensors to enhance the spatio-temporal resolution of the PM2.5 pollution map.

​

US Patent:
Yi, W. Y, Leung, K. S., & Leung, Y. (2020). U.S. Patent No. 10,533,981. Washington, DC: U.S. Patent and Trademark Office.

We have been developing a personalized healthcare system, including a wearable ECG device for acquiring users’ electrocardiography (ECG) data, a mobile application and web interface for data visualization, and a ResNet base Atrial Fibrillation (AF) detector. The system is developed for people to gather their own health data from different sensors, such as smart watch, blood pressure sensor, heart rate sensor and ECG sensor.

​

This system provides a convenient and light way for users to keep track of their health condition via wearable devices and smart phones. Also, the system enables the health advisors to monitor and analyze their clients’ ECG information remotely through a web interface.

​

This is an on-going research and we have refined the design of the wearable ECG device and made five of them for trial use in Prince of Wales Hospital; secondly we refined the mobile application for better visualization and continuous monitoring; thirdly we improved the AF detector to achieve 84.37% mean F1 measure.

The major goals of the system and its online application are to develop an online application to allow user to annotate and query the taxonomy learning system which is based on WordNet; and to enable life-long learning of the system for taxonomy extension using iterative feedback from crowdsourced user and Natural Language Progressing (NLP) experts.

​

We have developed the first prototype of the system and finished the alpha testing stage. With multiple loops of human feedback and system self-learning, the system showed substantial improvements in terms of human explainability and prediction accuracy in solving lexical relationship prediction task. We planned to further improve the system by adding neural network models and integrating more knowledge base in the future.

We are currently collaborating with the Department of Psychology, CUHK and National Cheng Kung University on developing a mindful flourishing app.

​

The main idea is to popularize culture of mindfulness in colleges and provide a mobile platform for students to promoting their mental well-being. The application contains primary mindfulness practices, real time mindfulness action and online information for mental health. Through daily use of the app, students’ daily record can be further deployed to research on how cultivation of mindfulness can contribute to reducing pressure and improving mental health.

​

The Mindful Flourishing system has been online for experimental data collection since the Android APP launch in April 2020 and has gained over 4000 test records from about 120 users. To further enlarge the scale of the experiment, the IOS application is under development. 

​

**This project is supported by the Health Care and Promotion Scheme.

The major goals of the Objects Tracking AI system using LiDAR Sensor are to develop a robust AI system to track human or other object’s movements such as old-people falling without any privacy issue (Lidar cannot review any face nor body shape information.); to identify a human or an object from LiDAR Raw data; to develop a fast and precise identification method to enable real-time tracking monitoring.

​

​There are many potential applications for this system, for example: security and safety monitoring of old people or patients with mental problems, part of the sensor on Smart Lamppost, improving the privacy issue compared with using cameras, and LiDAR operation without any visible light source. There are several outstanding features in our system, for example: LiDAR Sensor and computer integration to build up a real-time and precise tracking system; a high-end LiDAR Sensor to ensure large sensing range and precise data collection; a generic Neural Network AI system to recognize multiple data format of LiDAR; and the AI system can deploy in a low power consumption computer system.

The major goals of the platform and its online application are to develop an online application to assess student behaviors and the outcomes of positive education intervention; and to enable real-time and near-real-time iterative tracking of and feedback to students to facilitate the design of effective teaching pedagogy for positive education.

​

The online applications contain several desirable features. It is a general all-in-one platform that is capable of real-time, near-real-time and long-term monitoring and tracking of the learning behavior and well-being of students in a school, which is very useful for developing and improving the curricula, practices, teaching and learning materials for positive education. Multimodal data will be automatically collected to facilitated long-term research on positive education.

​

We have developed a prototype of the online application and we will perform a pilot test on July 2020 to test and evaluate the application. The pilot test will be held in a Form 1 orientation camp at a secondary school with 70 students. We will create a positive education survey using our application and the students will answer the survey using the iPads from the school. The research team and the teachers will then be able to view the data collected from the students to understand student’s behaviour.

​

**This research is supported by the Bei Shan Tang Foundation (Dec 2018 - Nov 2020).