Comprehensive Modeling of Beam Propagation in Multimode Fiber and Experimental Validation (Year 2) Team 23044
PROJECT GOAL Develop a graphical user interface (GUI)-enabled, experimentally verified optical fiber beam propagation software package that accurately models how fiber bending, twisting and ambient temperature impact performance. Optical fibers are waveguide conduits that transport light. Due to their unique optical confinement mechanisms, optical fibers can only carry a finite number of modes, or electromagnetic (EM) field distributions. Although the physics of straight fibers is well understood, many applications require the waveguides to perform in more complex environments, which significantly alter the behavior of the modes and, therefore, the guided field. Engineers can compensate for these environmental perturbations, but only with accurate computational modeling. This team based its approach on the software simulation package from last year’s Team 22041, improving the package usability and fiber bending accuracy with verification through a rigorous experimental study. The simulation can be accessed through a fully functioning GUI that allows the user to propagate Gaussian, flat top, annular and custom EM fields through step index or photonic crystal fibers with arbitrary bending/twisting geometries and ambient temperatures. During the simulation, the GUI displays the fiber layout, propagation loss, and real-time updates of the EM field as it propagates through the fiber, providing the user with enhanced analysis and streamlined fiber-based optical design.
TEAM MEMBERS Francisco Javier Flowers, Optical Sciences & Engineering Oscar Hsueh, Electrical & Computer Engineering Atkin David Hyatt, Optical Sciences & Engineering Lauren McCaffrey, Optical Sciences & Engineering Oliver Wu, Optical Sciences & Engineering COLLEGE MENTOR Mike Nofziger SPONSOR ADVISOR Tao Chen
Vision-Based Agricultural Implement Awareness Team 23046
PROJECT GOAL Using various sensors, identify and track the position of any implement attached to an autonomous tractor. Adding the capabilities of implement recognition and positioning to a tractor dramatically increases the scope of what tasks its autonomous system can perform. This design serves as a cost-saving alternative to adding automation equipment to each of the several implements that a single tractor pulls. The system, which is completely powered by the tractor itself and can withstand agricultural conditions, sits on top of the tractor cab to visually recognize and track the towed implement. It uses an OAK-D-PoE camera with machine learning models to perform implement detection and identification and a Livox Mid-40 sensor with lidar, or Light Detection and Ranging, and traditional algorithms to accomplish positioning relative to the tractor. All of the visual information is compiled and calculated on an NVIDIA Jetson in conjunction with the OAK-D, with on-board processing capabilities. The Oak-D performs the Neural Network computing on the camera itself, while the NVIDIA Jetson processes the outputs from the neural network and lidar data as well as the overall system code.
TEAM MEMBERS Gavin M Caldwell, Electrical & Computer Engineering Jessica S Grove, Industrial Engineering Brett Miller, Biosystems Engineering Kees Passon, Optical Sciences & Engineering Everett Schafer, Optical Sciences & Engineering Howard James Yawit, Electrical & Computer Engineering COLLEGE MENTOR James Sweetman SPONSOR ADVISOR Darcy Cook
The result is a complex network of embedded software and hardware integrated to accurately identify and locate implements, effectively advancing the future of automation in agriculture. PROJECT DESCRIPTIONS
27