School of Engineering Department of Electronic and Computer Engineering 147 A Swarm of UAVs Supervisor: SHEN Shaojie / ECE Student: YU Kaiwen / COGBM Course: UROP 1100, Summer Autonomous discovery stands as a critical function across a spectrum of unmanned aerial vehicle (UAV) uses, providing the basis for subsequent UAV path planning and actions. Two primary challenges exist: efficiency and accuracy. Firstly, numerous existing approaches face challenges related to ineffective global coverage, cautious trajectory plotting, and infrequent decision-making, resulting in subpar exploration rates. Secondly, a significant issue encountered by numerous approaches is localization drift, resulting in substantial distortion of the reconstructed map. In this project, I analyzed two related papers "FUEL: Fast UAV Exploration using Incremental Frontier Structure and Hierarchical Planning" and "Exploration with Global Consistency Using Real-Time Reintegration and Active Loop Closure". In the future, I aim to develop a novel algorithm that integrates good methods from these two papers and strikes a fine balance between efficiency and accuracy for UAV exploration. While the detailed implementation phase has not commenced, this report outlines our journey thus far, detailing the initial analysis and future plans for enhancing the quality of autonomous exploration. AI Enhanced Smart Design, Manufacturing and Robot Supervisor: SHEN Yajing / ECE Student: QIN Zhengyan Lambo / CPEG Course: UROP 1100, Fall UROP 2100, Spring My research task was to create a computer vision system that can align a robot gripper to a cable plug slot of a raspberry pi. The goal of this project is to align the robot gripper roughly with the cable slot, after which our tactile sensing deep learning system will be used to precisely plug in the cable and detect if the insertion was successful. This project is unfinished, but work have been done to try different approaches to find a solution for computer vision alignment. The methods I have tried include template matching, You Only Look Once (YOLO), and Segment Anything Model (SAM). More and better training data is needed to make the algorithms more accurate. Currently SAM shows the most potential for cable slot detection, but lack the ability to locate the target by itself. AI Enhanced Smart Design, Manufacturing and Robot Supervisor: SHEN Yajing / ECE Student: WANG Penghao / ELEC Course: UROP 1100, Fall As a student majoring in Electronic and Computer Engineering (ECE), I have always been dreaming of having a research experience in the ECE-related area. Fortunately, my advisor Professor Shen Yajing is offering UROP opportunities, so I decided to apply for it. My UROP1100K project starting from this semester is mainly about 3D-printing flexible circuits. With the help of Professor Shen Yajing and his Ph.D. student Guo Dong, I am able to gain a deeper insight into the 3D-printing industry and construct a 3D printer with another ECE student Huang Kaifeng (He did not enroll in UROP. He is helping me as a volunteer.) Overall, the research experiences this semester are challenging but fruitful.
RkJQdWJsaXNoZXIy NDk5Njg=