2024-2025 Teams

To advance power system studies, researchers develop synthetic grid models to mirror the real-world power networks. They utilize publicly available data and the laws of physics to generate realistic models that emulate the performance of the actual grid without releasing sensitive network data. For years, the synthetic test cases have been generated for most parts of Northern America and Europe with little to no focus on other parts of the world.

For this project, we would attempt to build a first of its kind synthetic test case for Ghana, a country in West Africa. Since most of the previous work in this field has been done for European and North American grids, we would investigate whether the same techniques applied can be used in Africa especially considering the vast differences in factors that affect grid operations. (E.g. climate, infrastructure investment, maintenance culture etc.). We culminate the program by developing a Matpower file representation of parts of that Ghanaian electricity grid and perform multiple power system operations to confirm its validity. A research paper would also be submitted to the 2025 IEEE PES Africa conference for publication.

This ORS project’s goal is to research and create unique waveform design and simple circuit schemes to increase read-ranges of radio-frequency identification (RFID) tags that operate battery-less. (How can we communicate at longer distances with low power on the order of micro-watts?)

RFID tags are being integrated in the growing Internet of Things to help wirelessly track objects, monitor environmental changes, etc. Specifically, passive RFID tags aim to be durable and inexpensive; thus, not relying on batteries to power their internal hardware or communicate with a tag reader. Even though these tags’ read-ranges are often constrained by how efficiently they can harvest energy from signals in their operating environment, unique schemes are being researched to allow these tags to operate at longer ranges. This project will allow the students to decide on and explore different ways to improve passive tags’ read-ranges through methods like designing and transmitting power-optimized waveforms, fabricating retrodirective tag structures, and more. This project is interesting because it challenges the researcher to figure out creative and simple ways to “do more with less” as tag designs must be kept simple due to their power constraints.

The project aims to advance mobile manipulation for indoor tasks by achieving more robust and reliable object recognition and collision-free grasping strategies. This project will develop an integrated system that combines the precise manipulation capabilities of a robotic arm with the mobility of a mobile base, enabling seamless execution of complex tasks in dynamic indoor environments. There will be a focus on creating sophisticated planning algorithms that incorporate both manipulation and base movement, ensuring efficient and safe task execution.

A key innovation will be the derivation of ego-centric world models, which provide the robot with a detailed, first-person perspective of its surroundings. These models will be integrated into compact world representations, facilitating better decision-making and planning. By doing so, the project aims to enhance the robot’s ability to navigate and manipulate objects within cluttered and unpredictable indoor spaces, ultimately improving the robot’s autonomy and effectiveness in assistive applications. The outcome will be a versatile and intelligent mobile manipulator capable of performing a wide range of assistive tasks, from fetching items to assisting with daily activities, significantly enhancing the quality of life for individuals in need of assistance.

Analog VLSI has lagged behind Digital VLSI because of a lack of tool infrastructure and circuits amenable to automation. Recently our lab has made strides in both directions, designing unique programmable analog standard cells to be used in a novel tool framework. The first set of standard cells was designed in the open-source Skywater 130nm process, but since then we have made advances in both standard cell design and tool infrastructure. Students in this project will use our new knowledge to modify the existing Skywater 130nm cell library using the open-source toolchain and then synthesize Analog VLSI circuits with our toolchain using the cell library they helped create.

This project aims to develop a neural receiver that outperforms conventional receivers by achieving lower bit error rates (BER) through the integration of 3GPP TDL/CDL models and site-specific ray-tracing channel models. By combining the statistical properties of standardized 3GPP models with the detailed environmental insights from ray-tracing simulations, we create a comprehensive channel modeling framework. This integrated approach provides rich and diverse training data for the neural network, enabling it to learn more robust signal processing techniques. The neural receiver is trained on this dataset to enhance its ability to accurately decode transmitted signals in complex and varied wireless communication environments.

The field of analog computing holds promise for generating significant power and area savings at equivalent process nodes. Yet before we can fully realize that vision the corresponding tools and inputs must be built up. Our lab has made progress on both these fronts by creating and testing analog standard cells across various process nodes while also designing a novel analog toolchain for programming a reconfigurable analog computing platform (Field Programmable Analog Array- FPAA).

This project consists of extending the work of the toolchain that will focus on the writing of code and testing of the FPAA specific programming. The goal is to harden the integration done by a previous generation of ORS students, update user facing libraries to support a range of circuit design use cases and finally program & test various analog computing circuits like matrix multiplication, amplitude detectors, softmax circuits and the like. Given sufficient progress, students can test their work by comparing some simple one layer or multi layer neural networks implemented in the analog domain.

Low frequency radio wave signals (1-500 kHz) are a useful tool to study the D-region of the upper atmosphere. Lightning strikes, solar flares, and longwave radio broadcasts have impacts on this region which can disturb or even disrupt global communication systems. To research and understand these impacts, the LF lab has a global array of low frequency radio receivers which collect and return data in real time for analysis. This system consists of a pair of antenna, an analog circuit system to condition and amplify these signals, and a data acquisition (DAQ) system which digitizes the data for storage and analysis. This DAQ takes in analog data from lightning strikes and converts it to a digital signal. The aim of this project is to update the DAQ design to lower cost and improve performance over the currently implemented design. The project will consist of hardware programming, software development, embedded system design, FPGA and microcontroller subprojects. Once a DAQ system is designed and validated, it will be implemented into thw low frequency radio receiver system.

Renewable resources such as solar and wind energy are being introduced to electric distribution grids at high rates; however, the behavior of these resources is inherently random, introducing uncertainty into computational grid models. This uncertainty arises both when ensuring that physical engineering constraints are satisfied and when we wish to validate the accuracy of computational grid models.

This project aims to investigate the effects of randomness on the power flow equations and improve grid state estimation by interweaving electric grid physics with emerging techniques from probability theory. We will develop methods to efficiently use uncertain measurement data to perform grid state estimation, focusing on optimizing smart meter data streams for accuracy and efficiency. The goal is to maximize the accuracy of grid state estimation while minimizing the number of sensors used and the frequency of the data samples taken.

Task 1: Develop a generative machine learning model to produce synthetic power system datasets that reflect uncertainty in renewable resource scenarios.

Task 2: Design a signal processing algorithm to strategically select power system sensor sampling rates while optimizing accuracy and efficiency.

Task 3: Analyze the theoretical properties of the sensor sampling rates and integrate these findings with the signal processing algorithm of Task 2.

Task 4: Implement and test the developments in Task 2 and Task 3 with the datasets developed in Task 1.