Simulations and Software
As an aspiring mechanical engineer with an emphasis on mechatronics, I have worked on several projects that incorporate programming and computer-aided engineering software. Featured below are a few of these projects. The technical tools that I have extensive experience with include SolidWorks, Matlab, COMSOL Multiphysics, Ansys Workbench, python, and C++.
Optimal Observer-Based Controller Design for the Inverted Double-Pendulum
UC Santa Barbara, ME 243A (Linear Systems I) - Fall 2023
Similar to the previous project on the portfolio (see below), this project presents an optimal controller design for a slightly more complicated system: the inverted double pendulum. The paper shown below features a similar engineering procedure, including system modelling, controllability studies, optimal controller design, observability studies, and optimal observer-based controller design. However, this project has a more theoretical foundation and the controller is only implemented in simulation. Several sections in the paper provide commentary on the design lessons learned from this exercise and how they would be applied to future hardware implementation.
System Identification and Optimal Control of an Inverted Pendulum on a Cart
UC Santa Barbara, ECE 147C (Control Systems Laboratory) - Spring 2023
For the final project of our control systems design laboratory, we designed an observer-based controller for the Quanser linear servo base unit with an inverted pendulum. The system features two encoders recording linear and angular position and a single actuator driving the cart on a rack. To achieve this goal, we first characterized the system using parametric system identification and analytical modeling. LQR/LQG controller design was first executed in MATLAB using the control toolbox, and fine-tuned in hardware implementation. The controller was implemented using simulink via Quanser amplifier and data acquisition boards.
The system matched all of our selected performance metrics, including a 5 degree angle tracking error, low-frequency disturbance rejection, and high-frequency noise rejection. The final report attached below includes a detailed overview of our analysis and the system's performance.
Modeling and Control of Apical Extension Robots Through Hybrid Dynamics
UC Santa Barbara, ECE229 (Hybrid Dynamics) - Winter 2023
Hybrid dynamics is a framework for modeling dynamical systems that exhibit both continuous-time and discrete-time behavior. As the final project for a grad-level course on this topic, I presented an attempt at modeling and controlling vine robots (see the design page for more info on this!) through this framework. The report below includes a brief introduction to hybrid dynamics, a model of the vine robot as a system with explicit logic modes, a brief stability treatment of the candidate controller, and computational simulations of the system.
Modeling of Growth and Retraction of Apical Extension (Vine) Robots
Apical Robotics Capstone Team, Santa Barbara - Fall 2022 to Present
Inspection of industrial piping systems is a significant engineering challenge. The objective of this project is to develop a vine robot (see below for information on this technology) for inspecting currently inaccessible pipes or other industrial equipment. As a short-term deliverable, we are focused on optical inspection of a pipe section 3' in diameter with a 120' long horizontal section followed by a 40' vertical section.
Blumenschein et. al. (linked below) outline an analytical model for predicting the required pressures to grow and retract vine robots. This information is critical in assessing the required burst pressure of the robot as well as selecting motors, fans, and hardware. I was responsible for applying an extended version of this model to the aforementioned inspection task. The analytical portion of the modeling was aided by my teammates and our advisor and the models were implemented in MATLAB. More information on the modeling is provided in our quarterly presentation, attached below.
Implementation of Feed-forward, Convolutional, and Recurrent Neural Networks in TensorFlow
UC Santa Barbara, ME225 (Research Topics: Machine Learning) - Winter 2023
This course presented deep learning as a tool for engineering modeling and exploration with an emphasis on both the statistical foundations of learning and implementation of neural networks in TensorFlow 1.x and 2. Tasks included non-linear data fitting, classification by feedforward networks, and image processing via CNNs - the PDF attached below is from a script concerned with the latest of those topics.
Motion Planning Software for Robotic Applications
UC Santa Barbara, ME179P (Robotics: Motion Planning and Kinematics) - Winter 2021
The first half of this robotics course focused on several fundamental motion planning schemes: sensor-based planning, planning via decomposition and search, and planning via sampling - including incremental tree-roadmap computation. The attached report provides an algorithm for sensor-based planning for an environment with multiple convex polygonal obstacles.
Solid Static Simulations of Limb Deformation
Galva Applied Science (PI: Dr. Nikolas Kastor, Advisor: Dr. Piers Echols-Jones) - Summer 2021 to Present
Under the supervision of my advisors, I worked on developing 2D, axisymmetric ANSYS Workbench simulations that model limb deformation under an applied cuff pressure. Perhaps the most significant challenge in developing this model is the implementation of appropriate boundary and contact conditions that accurately reflect the limb's physiological mechanical constraints. The simulations are guided by available literature in this area.
Attached below, a sample plot of axial stress is presented from one of the custom limb geometries.
Multiphysics Heat Transfer Simulations for a Novel Haptic Actuator
UC Santa Barbara, RE Touch Lab (PI: Dr. Yon Visell, Advisor: Dr. Nikolas Kastor) - Fall 2020 to Present
The RE Touch Lab studies haptics (application of technology that stimulates a sense of touch and motion) and soft robotics. One of the ongoing endeavors in the lab is the development of a novel haptic actuator, which is able to operate in a wide range of frequencies while maintaining a small size.
Thermal failure is a limiting factor for many compact electromagnetic actuators. Subsequently, my work has focused on carrying out thermal simulations using COMSOL multiphysics to guide design decisions from a thermodynamics perspective and evaluate the device's thermal performance.
The button below is a link to the final publication incorporating this work.
Finite Difference Simulations for a Two-Dimensional Advection-Diffusion Cooling System
UC Santa Barbara, ME17 (Mathematics of Engineering) - Spring 2021
This simulation was the culmination of a set of numerical methods developed within the class for mathematics of engineering. Topics of interest included ordinary differential equations, partial differential equations, and eigenvalue problems.
This project in MATLAB simulates the transient progression of a cooling system in 2 dimensions. The simulation accounts for advection (net transport of the coolant fluid) and diffusion (of heat within the fluid). Finally, Robin boundary conditions are used to simulate convective cooling. A complete report of the project can be found below.
Thermal Simulations for a Phosphor Light-Conversion Platform
Fluency Lighting Technologies, Santa Barbara (PI: Dr. Kristin Denault, Advisor: Marc Viray) - Summer 2020
My project with Fluency Lighting Technologies over the summer of 2020 focused on evaluating the effective thermal properties of Fluency's proprietary phosphor converter. In specific, my project examined the impact of several imperfections and manufacturing flaws on heat transfer within the chip. My contributions included developing 3D models of these imperfections with varying degrees of complexity, performing steady-state and transient thermal simulations using these models, and interpreting the results of these simulations to inform future design.
The poster below includes a more comprehensive overview of the project.
Deep Learning, Reinforced Deep Learning, and Monte Carlo Algorithm Implementations
UC Santa Barbara, ENGR10H (Introduction to Programming for Engineers) - Spring 2020
The honors section for the introduction to programming course included the implementation of several fundamental principles in machine learning. These included the basics of image processing, several image-recognition neural networks of varying complexity, and a final project that aimed to complete a custom game using artificial intelligence.
Robotics Club C++ Projects
UC Santa Barbara Robotics Club - Fall 2020
Over the course of the fall, the UCSB Robotics Software team developed several obstacle detection and motion planning programs in C++. The final project included a simulation of that year's Vex U competition. Featured below is the Github repo that includes the final program.
Implementation of Control Systems to Sustain Autonomous Robot Motion
Summer Institute in Mathematics and Science (SIMS) - Summer 2019
With the supervision of a graduate student mentor, our group developed an elementary PID controller to accomplish a pursuit and evasion task using retrofitted Roomba robots. This included breaking down the system dynamics using a linear-time-invariant model and subsequently designing and tuning a controller.