Jueun Kwon

We developed the vision module, designed the mechanical putter end-effector, and created the ROS2 package for Apex Putter, a mini-golf system using a Franka Panda arm.

The vision system integrates an AprilTag-based robot base registration and a YOLOv8-trained model for ball and hole detection. The mechanical system features a custom end-effector designed for optimal putting accuracy. Motion planning is handled via ROS2 and MoveIt2, ensuring precise force application for successful putts.

The GitHub repo can be found here.

Tiantian and I developed Babysub, a thruster-propelled, small-scale autonomous underwater swarm vehicle, focusing on affordability and customizability for broader access to underwater exploration.

The experimental outcomes in agent tracking and following validate Babysub’s effectiveness in essential underwater tasks. Additionally, Babysub’s performance in a controlled environment has laid the groundwork for its potential use in more complex, real-world scenarios, particularly in swarm-based applications.

Project report can be found here.

I was a mechanical design engineering intern on silicon engineering group’s silicon validation hardware team at Apple.

During the internship, I proposed and developed a system to measure the current carrying capacity of SOC sockets. I designed and built a test fixture that synchronously measures the current, force, and temperature of individual pogo-pins. In addition, I orchestrated vendors to advance and manufacture custom designed parts for the fixture system. Lastly, I collected, analyzed and interpreted test data on the test fixture I built to verify vendor’s pogo-pin specs.

From September 2022 to May 2023, my mechanical subteam co-lead and I designed and manufactured the complete frame structure of Cornell University Autonomous Underwater (CUAUV)’s, from scratch. The frame is the backbone of the autonomous underwater vehicle’s structure, securing all enclosures and additional components around the main enclosure, the upper hull pressure vessel (UHPV). The frame’s component placement is largely responsible for how the submarine interacts with its environment.

Component placement and mounting affects buoyancy, the functionality of the enclosures and components, and the protection of these components from external impact damage. In addition to manufacturability and sufficient safety factors from typical loading, the frame design considers the specific restrictions and objectives of each enclosure and component. These specific objectives are balanced with the broader objectives of the vehicle, such as reducing pitch and roll, maximizing enclosure accessibility, and organizing enclosures in a spatially efficient manner. The frame design alone dictates vehicle layout and that made designing the frame one of the hardest technical challenges I faced thus far. Through multiple design reviews, alumni feedback, and sheer hours on Solidworks, the frame was able to be designed. At the end of the project, the submarine was trimmed with less than 3 degrees of tilt on every axis without addition floats (foam) or additions weights.

Through a class called fasts robots, I implemented PID control, Kalman filters, localization, and path planning on a robot car via Bluetooth. The lab reports can be found here: https://jueunkwon.github.io/

At Cornell’s 24-hour Make-a-thon, two of my friends and I built a smart trashcan that can recognize and sort 6 different types of waste to sustainability reduce landfill waste and encourage recycling, winning 3rd place.

On the software side, we researched various datasets for recyclable image classification and decided to use a combination of TrashNet (from Github) and Waste Classification Data (from Kaggle). We used a pretrained VGG-16 deep convolutional neural network, with our own custom classification head. Our model achieved an accuracy of 85% overall. We did some fine tuning and biased predictions towards ‘trash’, achieving 95% accuracy on the trash category in the end.

On the hardware side, with limited access to hardware and fabrication, we were able to make a concept trashcan that displays the trashcan’s overall goal. The trashcan is fabricated with 3d printed, laser cut parts, and fasteners. The trashcan 6 different waste bins with one neutral position. When the trashcan recognizes what type of trash it is, the servo turns to open up the corresponding trash bin (trash, cardboard, glass, metal, paper, and plastic).

Further software and hardware implementations can be found in the video and the presentation.

Team members: Jueun Kwon, Tiantian Li, Raymond Chang

I was a mechanical engineering intern for ocean hardware design on project Tidal at X, the moonshot factory (formerly known as Google X).

During my internship, I led module level reliability testing to downselect inserts and gaskets for a watertight seal in an ocean environment, including a presentation to senior leadership. I also worked cross functionally with manufacturing and product design to incorporate learnings from reliability failures. In addition, I designed and tested fixtures for monotonic tensile testing and chamber rigs for use in extreme ocean environments.

I designed a testing set up to measure force from the CO2 canister by a force/torque sensor inside the thermal vacuum chamber for Cislunar Explorers. The designed testing set up measures the force generated by the ACS nozzle. This allows us to find the total thrust generated by the CO2 canister through the ACS nozzle and that value can be used to calculate the total reorientation capacity of the CO2 canister.

At Cornell University Autonomous Underwater Vehicle, I designed and conducted analysis of new torpedoes and droppers systems for the autonomous underwater vehicle using Solidworks. I individually modeled the torpedo and dropper launch systems, which was then further refined through four design reviews with senior members of the team to validate my design to meet all the technical and manufacturing specifications. My design choices were able to reduce a lot of weight, space, and fuel (compressed air). This semester, I am manufacturing my design to be put on the competition vehicle.

Technical documentation can be found [here]

At Autonomous Systems Lab (ASL) at Cornell University, I worked on simulating aerial navigation using the Airsim simulation library in the Unreal 3D game engine. AirSim (Aerial Informatics and Robotics Simulation) is open source robotics simulation platform and Unreal Engine is a simulation game engine. Using AirSim, I set up a photorealistic simulation environment of a drone flying through a forest. I controlled the drone manually using an Xbox controller and saved images from a 1st person view with corresponding logs that map to its time stamp, position, and image name. In addition, I Iterated upon getting depth images (ground truth) and two camera images from the same horizontal axis to later build and compare depth images.

During my internship at NYU Agile Robotics and Perception Lab, I worked on drone designs and wrote C++ controller code for research on swarm drones for autonomous package delivery. I designed and 3D printed drone parts, camera enclosures, frames, and attachment links for drones.

I was acknowledged on Cooperative Transportation of Cable Suspended Payloads With MAVs Using Monocular Vision and Inertial Sensing.

Mohamed Bin Zayed International Robotics Challenge (MBZIRC) is an international robotics competition held in Abu Dhabi, UAE, every two years. I was involved in challenge #2 (bricks) where team of 3 UAVs and 1 UGV has to build a wall using colored breaks. I was responsible for the hardware connection between Jackal Unmanned Ground Vehicle and Kinova Gen3. In addition, I worked on the end-effector design and assisting with testing by re-creating competition game pieces.

Our collegiate team of NYU, UPenn, and Czech Tech University won the Grand Challenge.

I was acknowledged on NYU’s news found [here].

This project was created for MAE 3780: Mechatronics. My individual project is to turn off my light switch and turn on a red LED at night when I yell or press a button from my bed.

In terms of sensing, I have a button and a microphone sensor. The button can be pressed by the bed and the microphone sensor can detect loud sound to activate the servo and the red LED. In terms of actuation, I have a 180 degree servo and a red LED. The 180 degree servo can turn to flick off the light switch. The red LED can light up to indicate that the servo is moving. In terms of mechanism, rack and spur gear is used to translate rotational motion into linear motion. As the servo turns, the gear is rotated, turning the rack, and ultimately pushing the light switch off.

Technical report can be found [here]

I designed and animated the “Planetary Gear with Four-bar Linkage (O1)” model from the Reuleaux Kinematic Mechanisms Collection using Fusion 360. The “Planetary Gear with Four-bar Linkage (O1)” represents elements of machine motion. Two of the four linkages represents a sun-planet motion. Through this motion, the path of the planet gear traces an epicycloid curve

I made a model of a hand Sanitizer that accurately represents a physical hand sanitizer in Fusion 360. I included components such as the bottle, the pipe, the pump, the pump top, the screw cap, the coil, the ball, and the cap.

The goal of battlebots is to create a 15 lb robot that can most efficiently immobilize another robot. Designing for Battlebots poses unique challenges: one has to work within a certain weight limit while thinking about various factors affecting offense and defense. I enjoy the freedom that comes with building and the thought process in making a robot mobile yet invulnerable.

The project came to a pause while starting to machine the battlebot in Aluminum due to COVID 19 and school becoming fully virtual.

During my time at the Virtual Embodiment Lab (VEL) at Cornell University, I wrote literature reviews, managed data using excel coding for interrater reliability, created virtual avatars, acted as a research confederate, and taught research participants and visitors how to use the technology.

In a particular research project, participants needed to be hands-free to perform tasks without holding the Oculus Touch Controllers. Utilizing the Oculus open source CAD models, I designed and 3D printed modifications for the Oculus touch controllers.

During my time at Bergen Makerspace, I manufactured, debugged, and analyzed designs for individual or corporate projects on Wednesday Open Nights. In addition, I taught how to use computer-aided design software (Autodesk Inventor, AutoCAD, Fusion 360, Cura) and shop machines (laser cutter, CNC, 3D printer).

I was the co-captain of Titanium Knights Robotics A (15887). I mentored and lead team members in the build and design process.

I was the Team Technician, Safety Captain, and part of the Pit-Crew for Titanium Knights Robotics (6203). I designed and built robots for the FIRST Robotics Competition.