About Me

Josemiliano Cohen

Josemiliano Cohen
B.S. Electrical and Computer Engineering, Minor in Robotics
The University of Texas at Austin, Class of 2027
Undergraduate Researcher, Autonomous Mobile Robotics Laboratory
Teaching Assistant and Peer Mentor, Texas Robotics FRI Autonomous Driving

j6cohen@gmail.com

Interests

Autonomous Navigation, Embedded Systems, Robot Hardware Design, Machine Learning for Control

I build the hardware and software that let robots operate on their own outside the lab. My work spans embedded power and compute design, sensor integration, and learning-based navigation policies, with an emphasis on platforms that survive real deployment — high-speed off-road vehicles, legged robots, and scaled autonomous cars. Alongside research, I lead an undergraduate autonomous driving stream where I teach students to take a robot from bare chassis to autonomous operation.

Research and Experience

Josemiliano Cohen at the Space Vehicle Mockup Facility, NASA Johnson Space Center Shuttle payload bay mockup at NASA Johnson Space Center Johnson Space Center, Houston

NASA Johnson Space Center

Robotics Software Engineering Intern · May – August 2026 · Houston, TX

  • Built a C++ and Python ROS2 perception stack fusing stereo camera and 128-beam LiDAR data for real-time terrain reconstruction, generating terramechanics datasets used to validate lunar rover mobility for Artemis missions.
  • Implemented a sensor fusion pipeline integrating a Fixposition Vision-RTK 2 with hardware-timestamped LiDAR synchronization, reducing visual odometry pose drift by roughly 90% in high-glare outdoor conditions.
  • Architected the full multi-sensor payload — Jetson AGX Orin, three ZED stereo cameras, and an Ouster OS0-128 — end to end, from hardware bring-up through ROS2 driver integration and deployment for field operation.
AMRL research vehicle AMRL vehicle build

Autonomous Mobile Robotics Laboratory (AMRL), UT Austin

Undergraduate Robotics Researcher · August 2025 – Present

  • Designed and fabricated a high-speed (70 mph) autonomous off-road ground vehicle equipped with NVIDIA Jetson Orin compute and Livox Mid-360 LiDAR.
  • Integrated precision motor controllers and sensor suites for robust, real-time perception and navigation.
  • Engineered a fully embedded, Linux-based computing payload for Boston Dynamics Spot with a hot-swappable, multi-source power distribution system.
  • Assisted with social navigation data collection and contributed to development of RMA policies for navigation in crowded, dynamic environments.
  • Used SSH-based remote workflows for development, testing, and deployment to achieve sustained, compute-independent autonomy.
RoboRacer platform

Texas Robotics, UT Austin

Teaching Assistant and Peer Mentor, FRI Autonomous Driving · December 2025 – Present

  • Lead the autonomous driving software research stream, mentoring 30 students in ROS2, Docker, SLAM, and navigation policy development through imitation and reinforcement learning.
  • Guide students through full-stack autonomy integration, debugging, and deployment across perception, controls, and software workflows on a custom vehicle platform.
  • Designed, built, and validated 18 1/10th-scale RoboRacer platforms, later adopted in a UT computer science course, integrating Jetson Orin, LiDAR, Pi camera, VESC controller, IMU, and a custom PCB power system with MOSFET switching, protection diodes, and buck converters.
  • Authored hardware assembly guides, test procedures, and platform documentation in MkDocs; a publication submission to Human-Robot Interaction (HRI) is in progress.
FTC Team 13948 robot

FTC Team 13948, Cathedral Robotics

Co-Founder and Vice-President · January 2020 – July 2023 · El Paso, TX

  • Co-founded the high school FIRST robotics team and helped establish its structure, training program, and technical direction.
  • Designed a TAPPS Regional Championship robot using laser-cut steel, extrusion, Raspberry Pi, ASIC motor controllers, and Java-based autonomous navigation with vision and encoders.
  • Taught myself Java to build the robot software stack, then taught it to teammates to support programming and autonomous development.

Projects

Robotics

Spot payload 3 photos — click to enlarge

Spot Computing and Power Payload

2025 – Present

A fully embedded, Linux-based computing payload for Boston Dynamics Spot, with a hot-swappable, multi-source power distribution system using MOSFET-based automatic switching between shore power, onboard batteries, and Spot's native rail. Schottky diodes provide reverse current protection. The design carries two independent rails: 19V for the Jetson Orin AGX and 12V for a Microsoft Azure Kinect, LiDAR, and a USB hub for LiDAR Ethernet — keeping compute and sensor loads isolated and stable.

On top of that payload I am developing a socially-aware autonomous navigation stack in ROS 2 that lets Spot move through crowded human environments. The ongoing research targets the frozen robot problem, where conventional planners stall in dense pedestrian flows because their collision avoidance is too conservative.

ROS 2, Linux, embedded systems, power distribution, Azure Kinect, LiDAR, SSH workflows

1/10 scale autonomous vehicle 2 photos — click to enlarge

1/10 Scale Autonomous Vehicle Platform

2025 – 2026

Designed, built, and tested 1/10th-scale autonomous vehicles as the teaching platform for the 30-student FRI Autonomous Driving course. The platform combines hardware design, embedded systems, computer vision, and machine learning on NVIDIA Jetson Orin, Hokuyo LiDAR, a Pi camera, VESC controller, and IMU, and is documented for publication at the Human-Robot Interaction conference.

I also mentor undergraduate researchers on multi-agent coordination for the RoboRacer fleet, guiding development of decentralized planning algorithms for cooperative autonomous driving.

Autonomous driving, machine learning, computer vision, embedded systems, hardware design

Autonomous off-road vehicle

High-Speed Autonomous Off-Road Vehicle

2025 – Present

A 70 mph autonomous off-road ground vehicle built around NVIDIA Jetson Orin and Livox Mid-360 LiDAR, integrating motor controllers and sensor suites for real-time perception and navigation. Learning-based navigation policies are developed and trained directly on the platform.

ROS2, NVIDIA Jetson Orin, LiDAR, machine learning

FTC robot prototype 4 photos — click to enlarge

FTC Robot Prototypes

2023

Two full prototypes of a FIRST Tech Challenge regionals robot for Team 13948, covering mechanical design, electrical systems, and autonomous Java programming.

Java, Raspberry Pi, CAD

Joystick-controlled robot car

Joystick-Controlled Robot Car

February 2026

An Arduino-based robot car built for ME 350R Robot Mechanism Design, integrating an L298N motor driver, DC motors, and a joystick for real-time directional control with smooth speed interpolation. Particular attention went to cable management, structural integrity, and chassis construction.

Arduino, C++, motor control, mechanical design

Miniature robot leg

Miniature Robot Leg

2025

A miniature robot leg designed and fabricated as a testbed for PyTorch-based locomotion research, used for training neural network controllers with reinforcement learning and biomimetic gaits.

PyTorch, reinforcement learning, 3D printing, motion control

Slider crank mechanism

Slider Crank Mechanism

2025

A slider crank mechanism designed and analyzed for Robot Mechanism Design, covering mechanical design, kinematic analysis, and CAD modeling of rotational-to-linear motion conversion.

Mechanical design, kinematics, CAD

Embedded Systems

Space Invaders on MSPM0

Space Invaders on the TI MSPM0G3507

November – December 2024

A playable Space Invaders written from scratch for the TI MSPM0 microcontroller, with two language options, a start menu, score tracking, and leaderboards. Includes custom sprites, DAC sound output, ADC potentiometer control, and an ST7735 LCD display.

C, MSPM0G3507, DAC, ADC, ST7735 LCD

Traffic light controller 2 photos and a video — click to enlarge

Traffic Light Controller on the TI MSPM0G3507

October 2024

A four-way intersection controller built on finite state machines and an array of structs to prioritize vehicular and pedestrian flow, validated with a logic analyzer and oscilloscope.

C, MSPM0G3507, finite state machines

Software

Excel output of the Gmail job extractor Example of the generated spreadsheet

Gmail Job Application Extractor

May – August 2025

A Java application that scans Gmail for job application email, extracts company, role, and location with a language model, and exports structured records to Excel, with Gmail API integration, filtering, and error handling.

Java, Gmail API, OpenAI API, Apache POI

eBay Live Arbitrage Assistant

May – August 2025

A system that monitors eBay Live auctions in real time to identify underpriced items with resale potential, using language and image models for product identification, the eBay API for comparable pricing, and profit margin models to set bid thresholds. It generated over five figures in resale revenue.

Java, eBay API, NLP, data analytics

Education

UT Austin Electrical and Computer Engineering

The University of Texas at Austin

B.S. Electrical and Computer Engineering, Minor in Robotics · August 2023 – May 2027

  • Technical core in embedded systems and computer architecture.
  • Texas Robotics (teaching assistant and peer mentor), UT SHPE (member), IEEE (member), Longhorn Boxing (board).

El Paso Community College

A.A. Multidisciplinary Studies · August 2020 – May 2023

Completed an associate's degree concurrently with high school, graduating with over 60 college credits.

Cathedral High School

High School Diploma, GPA 4.4 · August 2019 – May 2023 · El Paso, TX

  • FIRST Robotics co-founder and vice-captain.
  • Chinese Club founder and vice-president.
  • Student body secretary and marketer.
  • National Science Honor Society service manager.
  • Varsity tennis, cross country, and band.

Activities