Ujin Kwon

Mechanical & Computer Engineering student building applied algorithms and vehicle software that move from numerical models to real hardware, field logs, and measurable results.

View selected projects Experience and background

Selected Work

Real-Vehicle Autonomy

Mapless Autonomous Parking connects FAST-LIO, live occupancy mapping, delay-aware tracking, and CAN-facing control on a Hyundai Ioniq.

View project

Measured Perception

Monoscale provides learning-free metric camera motion and dense occupancy, backed by CARLA evaluation and ablations.

View project

Deployable Numerics

ballistic-solver packages nonlinear interception into C++, Python, game-engine bindings, and an ARM64 hardware path.

View project

Field-to-Desktop Telemetry

A Raspberry Pi GNSS/RTK runtime feeds a PySide6 workbench for aligned laps, plots, maps, replay, and reports.

Runtime Analysis GUI

Evidence at a Glance

• Physical validation: real-vehicle parking and a Rock 5B + STM32 solver integration.
• Measured software: solver benchmarks, perception ablations, telemetry tests, and reproducible evaluation paths.
• Open source: three reviewed and merged Autoware Universe contributions.
• Research and operations: HuVILab autonomous-driving team lead and Luxon Racing Team telemetry collaboration.

Capability map Open-source work