Porting and optimizing machine learning models, including the Vision Transformer (ViT) model, CNN, and RNN onto FPGA platforms for enhanced performance.Utilizing the KR260 board and conducting emulation on the ZCU102 for comprehensive testing and validation.Focused on improving AI acceleration and optimizing resource usage for PyTorch models on hardware platforms, pushing the boundaries of real-time machine learning applications.

This research project involves the design and implementation of a custom FPGA video splicer for a room-sized 360-degree stereoscopic XR display, specifically a CAVE (Cave Automatic Virtual Environment) system utilizing LC shutter glasses. The splicer requires precise synchronization and signal integrity management to ensure the video is presented seamlessly, enhancing the immersive experience in XR environments. This endeavor contributes to advancements in digital hardware design, video signal… Show more This research project involves the design and implementation of a custom FPGA video splicer for a room-sized 360-degree stereoscopic XR display, specifically a CAVE (Cave Automatic Virtual Environment) system utilizing LC shutter glasses. The splicer requires precise synchronization and signal integrity management to ensure the video is presented seamlessly, enhancing the immersive experience in XR environments. This endeavor contributes to advancements in digital hardware design, video signal processing, and FPGA programming.The project utilized VDMA (Video Direct Memory Access) and AXI (Advanced eXtensible Interface) protocols, with interrupts to manage real-time events and ensure efficient operation. The system was designed and implemented using Vivado on a Zynq 7000 SoC on a Zybo Z7 board, leveraging its capabilities for precise control and synchronization of the video display. Building on this foundation, we are expanding the project to a larger scale using the ZCU102 and ZCU106 boards, which feature the Zynq UltraScale+ MPSoC. This upgrade will enable higher resolution inputs, more complex video processing, and enhanced I/O flexibility, driving further innovations in XR display technologies. Show less

The project focused on mapping crash data from I-795 in Maryland using MapBox and a JavaScript API. The map displayed different colors to indicate areas with varying levels of accident severity, offering a clear visual representation of high-risk zones along the interstate. The results were presented to a group of managers, providing valuable insights to guide decision-making and improve safety measures. This project demonstrated the effective use of data analysis and visualization techniques… Show more The project focused on mapping crash data from I-795 in Maryland using MapBox and a JavaScript API. The map displayed different colors to indicate areas with varying levels of accident severity, offering a clear visual representation of high-risk zones along the interstate. The results were presented to a group of managers, providing valuable insights to guide decision-making and improve safety measures. This project demonstrated the effective use of data analysis and visualization techniques to communicate complex information in a straightforward and actionable manner. Show less