Po-Hsun Chang
MS Robotics Student @ University of Michigan
I am passionate about building robotics that drives the future way of living. Currently I'm workiing as a graduate student research assistant on semantic navigation integrating LLM/VLM reasoning, semantic mapping, scene graph generation and mapping. Let's connect if you are curious about my work!
Skills & Technologies
Programming
Hardware
Tools
Education
My academic background and qualifications.
M.S. in Electrical and Computer Engineering: Robotics
Aug. 2024 - May 2026
University of Michigan - Ann Arbor
Specializing in Mobile Robotics and Computer Vision
Key Achievements:
- Top 3 in Robotics System Lab final competition
B.S. in Electrical Engineering
Sep. 2019 - Jun. 2023
National Cheng Kung University
Focused on SatCom, DSP and geospatial intelligence
Key Achievements:
- Received an honorable mention in 2022 Undergraduate Project Achievement Competition
- Achieved 2nd place in the project competition for the Data Engineering for Smart Cities course
- Top 10 Finalist Award in the 2022 RunSpace Innovation Challenge
- Published "A Satcom On-the-Move Phased-Array Antenna Tracking Algorithm on Robot Operating System." at ICCAS 2023 (pp. 1060-1065)
Professional Certifications
- Coursera: "Computational Motion Planning, Aerial Robotics"
- Udemy: "Autonomous Robots: Path Planning, Model Predictive Control, Localization"
Work Experience
My professional journey in robotics engineering.
Graduate Student Research Assistant
Aug. 2025 - Present
University of Michigan CURLY
- Working on development of semantic navigation algorithm for Spot
- Implementing LLM/VLM/VLA-based reasoning, open vocabulary detection, semantic mapping, scene graph generation, and planning in Isaac Sim
Machine Vision Development Intern
May 2025 - Aug. 2025
Tesla Inc.
- Deployed a height measurement program with 3.5x improved repeatability to production lines for cross-functional teams to analyze correlation between plunge force to avoid stuck in tester
- Collaborated with cross-functional teams to design 3D-based inspection systems utilizing OCR to prevent $4.5M material loss and 5-day downtime by ensuring robustness to surface residue and lighting variations
- Trained a 97% accuracy defect classification model, reducing manual inspection and empowering root cause analysis
- Enhanced 8% defect detection accuracy by implementing precise color calibration, reducing 53% color difference
- Built and maintained ML pipelines to automate model training and validation, improving efficiency and reproducibility
Research Assistant
Jul. 2023 - Sep. 2023
National Cheng Kung University Space Lab
- Software Defined Radio: Constructed Cross Ambiguity Function on signals of AIS to enhance navigation safety
- Signal Synchronization: Modified VHDL code of FPGA on LimeSDR, making GPS signal trigger data recording
- Digital Signal Processing: Implemented threshold method to detect prominent signals in raw IQ data
Undergraduate Student Research Assistant
Jan. 2022 - Jun. 2023
National Cheng Kung University Space Lab
- Led a team of 5 and established a modular ROS framework for user-selected sensor in satellite communication on-the-move antenna tracking, validated performance with TMYTEK mmWave kit, and presented at ICCAS 2023
- Achieved accurate attitude estimation through incorporating GNSS and reducing IMU drift error by performing sensor fusion of accelerometer and magnetometer data utilizing the Madgwick algorithm and Kalman filter
- Built a GTK and RViz-based user interface for real-time visualization of satellite pointing angles, significantly improving operational efficiency for antenna tracking systems
Projects
A showcase of my technical projects and engineering work.