Basic skills: Python

Preferred skills: Control and learning theory

Data: Simulation and real experimental data for a cart-pole system

Model: GPT-2 transformer

Number of positions available:2

References: Smith et al. 2026 (under review, email for PDF), Celestini et al. 2024, Open AI Gymnasium

Brief description: In this project, the student(s) will validate the safety performance of controlling an inverted pendulum on a cart, both in simulation and in practice using experiments. GPT-2.5 transformer neural networks can learn to control the cart-pole system using in-context examples. However, we do not yet know how safe or stable these controllers can be in practice. The student(s) on this project will implement the GPT-based in-context controller of the cart-pole system and test the performance in simulation, generate 3D simulations using Open AI Gymansium to confirm the performance, implement the controller framework with Quanser experimental setup for real-world testing, and validate the stability performance.

Student(s):

Basic skills: Python

Preferred skills: Machine learning

Data: Synthetic data from ODE models

Model: Neural networks for control

Number of positions available:1

References: Asikis et al. (code repo), Abu et al. 2023.

Brief description:The goal of this project is to study the effect of different cost functions on training neural networks for control tasks. Specifically, the underlying hypothesis is that neural networks do not model dynamical system stability in canonical training pipelines. Thereofre, in this project, the student(s) will reproduce Asikis et al.'s pipeline for training neural networks to perform control tasks and then compare their performance by changing the training strategies to explicitly include stability encoding. The stability can be encoded by formally defining Lyapunov-based cost functions such that stability properties are satisfied at all times. The expected outcome of the project is to show the neural network performance comparison in controlling dynamical systems.

Student(s):

Basic skills: Python

Preferred skills: Control theory

Data:--

Model:ODE

Number of positions available:1

References: Pandey 2025, Kumar et al. 2022, Khammash 2022, Zhang et al. 2025 arXiv:2506.10866

Brief description:In optoelectronic control, certain key parameters are important to achieve precise control while another set of parameters must remain unchanged. In this project, the student(s) will use models of optoelectronic control that include these parameters and then evaluate the robustness to uncertainties in each of these parameters by using a recently developed sensitivity-based method. The expected outcome of this project is a method to evaluate the robustness in optoelectronic and epigenetic control models. Additionally, this project will lead to new results on choosing best control strategies to be used in these applications.

Student(s):

Basic skills: Python

Preferred skills: Sympy

Data: --

Model: Ordinary differential equation models

Number of positions available:2

References:AutoReduce

Brief description:

Dimensionality reduction is a crucial technique in analyzing complex systems modeled by differential equations. This project aims to develop a robust validation framework for reduced-order models using AutoReduce. The student will work on designing a comprehensive set of unit tests using Sympy and integrating them into an automated validation pipeline that checks the accuracy and stability of reduced models against full-scale systems. Specifically the expected outcomes of this project are to: (1) Identify/design canonical unit tests for dimensionality reduction and build with sympy (using AutoReduce), (2) Develop a formal validation framework for dimensionality reduction with ODE models and their reduced versions, considering error and robustness metrics (Pandey et al. 2024).

Student(s):

Basic skills: Python

Preferred skills: Linear analysis

Data: --

Model:CRNs from BioCRNpyler

Number of positions available: 1

References: Pacti and AutoReduce

Brief description:Many engineering and biological systems are modeled using differential equations, but manually deriving formal specifications is time-consuming and error-prone. This project will use large language models to automatically generate system specifications in JSON format that can be verified using Pacti. The student will also integrate AutoReduce to analyze the validity and efficiency of these generated specifications through symbolic model reduction techniques. The expected outcome of the project is to develop JSONs using code-generating LLMs for Pacti and verify by integrating AutoReduce for symbolic dimensionality reduction.

Student(s):

Basic skills:Python and MATLAB

Preferred skills: Differential equations

Data:MATLAB models and simulations

Model:Physical system models

Number of positions available:1

References: The python-control package

Brief description: Education in upper division engineering courses often relies on MATLAB-based simulations. However, there is a strong push toward teaching the Python programming language in CS1 courses for engineering students. To leverage the stronger Python preparation among students and to prepare them better for the workforce (where Python is increasingly the preferred programming language of choice), it is imperative that we re-design the upper division activities and simulations in Python. The goal of this project is to take the first step in this direction, with a re-design of control systems examples and simulations in Python. This effort will contribute open-source educational libraries that can be adapted in formal and informal education settings easily.

Student(s):

Basic skills: Python

Preferred skills: CARLA/Unity or other backend simulation

Data: Waymo Open dataset and its variant, the WOMD Reasoning dataset.

Model: Llama 3b

Number of positions available: 3

References: Wei et al. arXiv 2024, Shi et al. arXiv 2022

Brief description: Large-language models are increasingly being explored for decision-making tasks in autonomous systems. This project will investigate how LLMs can be integrated into self-driving car navigation frameworks by using large-scale datasets such as Waymo Open. The project will focus on three key areas: designing a navigation policy, training on real-world trajectory data, and evaluating performance using reinforcement learning in CARLA or Unity simulations.

Student(s):

Basic skills: Python

Preferred skills: AI models in Python

Data: Student data from UC Merced

Model: LLama, GPT

Number of positions available: 1

References: Frias et al. 2025

Brief description: With the increasing use of AI in grading, integrating an autograder with platforms like Gradescope can streamline the evaluation process. This project will develop an AI-enhanced grading system that allows automated feedback generation using LLMs. The goal is to enable automated unit test generation, adaptive scoring, and integration into existing university assessment workflows. The student will work with Python, AI models, and automation tools.

Integrating a Python runtime environment to facilitate code execution and testing with Gradescope. The first (easy) step could be to have the tests be manual and combine the unit test grading with LLM grading to have a final grading score. The second step would be to generate the unit tests using an LLM and grading the code based on these unit tests. Finetune a model to create a “general unit test” creator. Use subprocess to handle inputs to the code. Use GPT-4 one-shot to generate the unit tests and then Llama to provide feedback and grade.

Student(s):