School of Engineering Department of Mechanical and Aerospace Engineering 161 AI-enabled Smart Pile Driving System Integrated with MEMS Sensors Supervisor: LEE Yi-Kuen / MAE Student: LAI Shing Kai / AE Course: UROP 1100, Spring Predicting the actual power consumption of a computer without relying on sensors is a complex task that has garnered significant interest in the field of machine learning. This interest stems from the increasing need to optimize energy efficiency in data centers and personal computing devices. As hardware sensors can introduce additional costs and complications, a machine learning approach provides a non-intrusive alternative to traditional power measurement techniques. By leveraging historical power usage data along with computer utilization metrics such as CPU load, memory usage, network activity, and disk operations, machine learning models can be trained to estimate power consumption with reasonable accuracy. Algorithms such as regression trees, neural networks, and support vector machines can map the relationships between these metrics and actual power usage to create a predictive model. Application of Artificial Intelligence to Enhance the Fluorescence Microscopy of Circulation Tumor Cells Captured by MEF Chips Supervisor: LEE Yi-Kuen / MAE Student: CHEANG Weng Io / COMP TRAN Chun Yui / COMP Course: UROP 1100, Fall UROP 2100, Spring UROP 1100, Fall UROP 2100, Spring The goal of this research is to explore the depth of applications of artificial intelligence to enhance the fluorescence microscopy of circulating tumor cells (CTCs) captured by microfluidic electrostatic field (MEF) chips. By trying out different deep-learning architectures, we aim to build a model that could speed up the detection of CTCs, while maintaining the high accuracy levels achieved by professional manual analysis. Ultimately, we hope the outcome of this research can be introduced and implemented into current fluorescence microscopy procedures, benefiting cancer diagnosis and monitoring workflows. This research has the potential to significantly improve the efficiency and accessibility of CTC-based cancer diagnostics by integrating an accurate, high-throughput AI-powered solution into standard clinical procedures. Development of MEMS Mirrors for Low-cost Lidar Used for Next-generation Self-driving Cars Supervisor: LEE Yi-Kuen / MAE Student: NEROTH Avram John / MECH Course: UROP 1100, Fall This progress report focuses on the development of Microelectromechanical Systems (MEMS) mirrors for low-cost LiDAR sensors used in next-generation self-driving cars. The report discusses the background of MEMS technology and its application in LiDAR sensors for autonomous vehicles. The objectives of the study include determining the damping coefficient of a resonant scanning micromirror using computational fluid dynamics (CFD) modelling and simulation. The report presents the theory behind the micromirror design, including air-damping effects, and describes the simulation model used in the study. Initial results and the next steps in the research are also discussed. The report aims to contribute to the understanding of airdamping characteristics in custom resonant scanning micromirrors for LiDAR applications.
RkJQdWJsaXNoZXIy NDk5Njg=