- Recipient of National Science Foundation grant appointment for capstone project
- Master’s Capstone Project: Created and integrated custom Python solutions to analyze treatment volume using 3D scans from ultrasound data
- Implemented functionality from third-party computer vision framework from scratch when unable to get time to meet with them to answer questions.
- Correlation between estimates and ground truth measurements ~0.95.
- Currently under review for IROS 2023 so I can't provide a link (for now), sorry!

- Recipient of National Science Foundation grant appointment for capstone project
- Master’s Capstone Project: Created and integrated custom Python solutions to analyze treatment volume using 3D scans from ultrasound data
• Wrote custom Python solution for ultrasound calibration pattern recognition and measurement
• Implemented software and hardware solutions for RS232-based end-effector remote control
- Explored porting existing x86-based software to ARM-based Nvidia Clara AGX dev kit
- Designed and executed numerous treatment simulations on agarose-based test articles (“phantoms”)
- Performed software validation for company’s first FDA De Novo submission
- Implemented custom solutions in C++-based third-party computer vision framework
- Collaborated with IT lead to create internal GitLab server and NAS backup volume

- Used Stable Diffusion to fortify an object-detection dataset with synthetic image variations.
- Achieved ~30% confidence increase for worst-performing images in baseline model
- Achieved ~10% confidence increase for already-well-performing images in baseline model
- The possibilities for this technique are endless, so exciting!
‍
‍

- Wrote program to search arbitrary directories for instances of some query image
- Results cached for fast future access
- Multithreaded
‍
‍

- Computer vision to measure (sub-mm precision) metrics of soft robot deformation (e.g.: inflection and min/max point positions, bend angle, arc length, etc)
- Interactive user interface for segmentation
- Future plans to implement deep-learning based segmentation in place of current color-based segmentation.
‍
‍

- Implemented pulse width modulation (PWM) and analog-to-digital converter (ADC) module functionality in TI C2000 microprocessor for cannon fire rate control
• Responsible for learning how to initialize those modules and understanding interactions between them
• Synchronized ADC measurements with PWM pulses to ensure valid readings of motor current
- Initialized and implemented I2C module -- TI C2000 microprocessor
• Communication with external EEPROM chip
- Implemented external crystal oscillator clock control on TI MSP430 microprocessor
All work done in C

- Created and updated tools for flight test telemetry data reduction
• For 30mm programmable airburst round and artillery guidance kit, respectively
• Wrote and validated user guide
• Work done in Python and MATLAB

- Gave presentation summarizing work and results to site upper management team
‍

- ROS, Turtlebot, Astra Pro camera, ArUco tags
- Room mapping with SLAM
- Autonomous navigation around room after mapping
- summoning to predetermined locations (and arbitrary ones with the use of one library we didn't have time to implement...)

-Trained a custom object-detection model with YOLOv3 on Darknet with no previous machine learning coursework or experience (aside from watching a lot of YouTube videos about it)
- Senior Capstone project with Toro
- Smart lawn infrastructure trimmer (trim grass around/over sprinklers, valve box covers, etc, but don't damage them)
- Implemented trained model on Nvidia Xavier running ROS Melodic.
‍

- Mbed controller
- state machine
- complicated tracks
(line breaks, crosses, obstacles to avoid, etc)
- 3d-printed chasis and wheels
- collect ping pong balls on the track with no moving parts

- Teensy 3.5 interfaced with absolute rotary and linear encoders
- learned what a PID controller is and how to make one for this project
- Recipient of Undergraduate Opportunity Program (UROP) grant for this work
- Published and Presented work at 2019 Design of Medical Devices Conference
‍