2022-2023 Teams

The earth’s atmosphere blocks most of the power that can be harvested from the sun; therefore, a clear alternative is to conduct further research into space solar power (SSP) using satellites and antennas. Larger transmission apertures are vulnerable to suffering catastrophic damage through environmental effects. Thus, this ORS project looks into a possible solution for SSP using disaggregated apertures for transmission (DAGATs) which involves having many small satellites in space coordinate together in a volumetric array. These structures would use antennas to transmit the power as a single signal to a surface station, which would provide increased stability and improved cost effectiveness. 

Our team will be scaling process nodes based on the IRDS roadmap and NCSU’s 3nm PDK to predict SRAM performance at 3nm.

We are using transmission measurements in the Ka-band to determine the attenuation constant of leaves. We aim to establish a relationship between the attenuation constant and the water content so that we can estimate the water content of a given leaf by measuring the attenuation constant. We will compare this estimate to the water content obtained from a conventional method that involves soaking and dehydrating the leaf. With our research, we want to contribute to estimating plant water content in a non-destructive fashion.  

Our project is to develop a low cost data acquisition system (DAQ) that processes analog signals produced by lighting in the ionosphere’s D-region. The DAQ comprises three parts: a GPS, an ADC, and a microcontroller. These devices will work in tandem to store the collected and time-stamped data.

Our group intends to develop a personalized thermal comfort prediction model that requires only a few thermal sensation votes from occupants to fit, using meta-learning. Under the framework of meta-learning, the learning of different individuals’ thermal preference models is modeled as different tasks sharing some similarities. We expect to gain personalized thermal comfort model with improved prediction performance over existing methods.

Inside of the 3D Systems Packaging Research Center our group’s research focuses on the optimization of the size of semiconductors using machine learning (both hardware and software components). Packaging size is important because a smaller package can allow for a smaller device as well as more room for new features of a device to be added.

In this project, we defined a car overtaking problem in a bidirectional lane. Considering the moving cars in both lanes, we will design an optimal controller for a car to realize the overtaking with minimal energy consumption and collision free. After that, we will consider the model with stochastic noise in the system and design a controller based on this scenario. Furthermore, we will try model-free method such as reinforcement learning to find an optimal trajectory to achieve the same objectives and compare the model-free method with model-based method.

Combining human teleoperation and autonomous navigation system is possible to improve the navigation performance through unknown environments. The project is building a vision-based navigation framework that can achieve the combination. With the help of human teleoperation, it is efficient to replan better path to the goal and avoid unsafe zones in advance.

Construction of a millimeter wave antenna measurement system using a Fanuc robot arm that enables precise movements needed to make planar, cylindrical, and spherical measurements of the antenna under test. Also, the conversion of near-field measurements to far-field in MATLAB.

The goal is to address challenges related to developing technology for passively powering sensors that will be needed to communicate with a base station or one another in smart city designs. This research project plans to reconstruct the traditional 1-to-2 Wilkinson Power Divider to become a 6-to-1 Wilkinson Power Combiner, design an RF detection circuit, and program a microcontroller to interface with the circuitry and provide feedback on beaming.

High-Level Synthesis (HLS) tools are becoming an increasingly popular methodology for circuit design thanks to their ability to convert annotated C++ code into low-level RTL circuit descriptions, enabling designers to create substantially more complex hardware designs with ease. HLS tools typically work by first compiling the C++ source code into LLVM IR (Low-Level Virtual Machine intermediate representation), then converting each instruction to RTL and scheduling it within a finite state machine (FSM). However, it can be difficult for designers to interpret the resulting RTL and scheduling information, making debugging difficult. Our ORS team is working on improving the interpretability of HLS synthesis output, including the generated LLVM IR and FSM schedules. This effort is in collaboration with Stefan Abi-Karam’s ORS team to create a unified web-based interface for an improved HLS developer experience.

Our team focuses on improving the development tools and workflow available to circuit designers using high-level synthesis (HLS) for digital circuits and FPGA design. Current vendor HLS tools lack many modern development features and conveniences that speed up development time and reduce the number of bugs in hardware designs. To address this, our team introduces a new open-source web-based report viewer as well as novel visualizations that any designer can access with minimal overhead from any server remotely. We also include new static analysis features looking at LLVM IR representations from the HLS flow in collaboration with Rishov Sarkar’s ORS team. We hope that this combination of improved and novel tooling makes HLS more accessible and easier for all students and designers.