The number of everyday smart devices, such as Nest and Samsung SmartThings, is projected to grow to the billions in the coming decade. Professor Shijia Pan and her students work to quantify the sensing quality in real-world applications, optimize the information these devices collect and model the changes in distributions of ubiquitous sensing data for accurate machine learning. They’ve recently been recognized for their efforts.
One of Pan’s scholarly papers about heterogeneous sensing was named Best Paper at this year’s Cyber-Physical Systems and Internet-of-Things Week’s Conference on Internet of Things Design and Implementation. She collaborated on the paper with other researchers from Carnegie Mellon and Stanford universities.
CPS-IoT Week is the premier event on cyber-physical systems and IoT research, and brings together seven conferences, multiple workshops, tutorials, various exhibitions from both industry and academia and the AutoCheckout Competition, designed to solve challenges in autonomous retail.
Pan, a member of the Department of Computer Science and Engineering, has been helping organize CPS-IoT Week and several other conferences since 2017, and said it’s an opportunity to ensure a good event that offers high-quality communication for the research community.
She also has been leading a workshop entitled “Data: Acquisition to Analysis” at the SenSys/BuildSys conference since 2018, focusing on enabling various cyber-physical systems’ dataset sharing and bringing in researchers to discuss how to regulate real-world dataset sharing for different cyber-physical sensing systems.
“It is not emphasized enough how precious large-scale, real-world datasets are for research,” Pan said. She also works with AiFi Inc., a startup that collaborates with universities to share such datasets to help the development of autonomous retail.
She used AiFi’s dataset this semester for a final team project in her class, Advanced Topics on Intelligent Systems. Rahul Sidramappa Hoskeri, who enrolled in the class and the competition, will extend his work on this topic to further his master’s degree thesis. And student Yue Zhang led a hybrid team with students from four other universities —Carnegie Mellon, Stanford, Tsinghua Berkeley Shenzhen Institute and Royal Melbourne Institute of Technology — that won third place out of eight competing teams in the AutoCheckout Competition. The team focused on location-aware multi-modal sensors in autonomous inventory-monitoring systems. Zhang has been a research specialist in Pan’s lab and has accepted her offer to join her lab as a Ph.D. student.
Additionally, Pan recently learned that her publication entitled “Step-Level Occupant Detection across Different Structures through Footstep-Induced Floor Vibration Using Model Transfer,” was named 2019’s Best Journal Paper by the American Society of Mechanical Engineers’ Structural Health Monitoring/ Nondestructive Evaluation Technical Committee.
The study takes sensing structural vibration for indoor human information in a new direction, repurposing traditional mechanical engineering knowledge for use in non-intrusive Internet of Things applications.
The award is given based on technical quality, originality of research and impact on the SHM/NDE community.
