陈冠廷

Sichuan University | Automation (Outstanding Engineer Program)

prof_pic.jpg

Phone: (+86) 13365983865

Email: cgtgetting@163.com

Chengdu, China

Hi, I am Guanting Chen, an undergraduate student in Automation (Outstanding Engineer Program), School of Electrical Engineering, Sichuan University.

My current interests focus on robotics, embodied AI, and visual perception.

Education

  • Sichuan University (Project 985 / Double First-Class), Automation, 2023.09 - Present
  • Weighted required-course GPA (first 5 semesters): 89.74
  • Rank (first 5 semesters): 11/164 (Top 7%)
  • Core courses: Automatic Control (93), Intro to Programming (94), Data Structures and Algorithms (92), Computer Organization and Design (96), Computer Networks and Communication (95)

Competitions

25th RoboCon Main Track (Vision Algorithm Team Leader)

  • Integrated and optimized relocalization in FAST-LIVO2 under unknown initial pose.
  • Built a cascaded pipeline with Teaser++ global registration and GICP local refinement, reaching robust real-time pose matching in large occlusion scenarios.

25th RoboCon Quadruped Track (Vision Algorithm Team Leader)

  • Adapted a self-designed 12-DOF quadruped to the HIMLoco framework and built RL training environments with terrain curriculum and dynamics constraints.
  • Independently developed Sim2Sim and Sim2Real transfer modules, including Isaac Gym to MuJoCo cross-validation and hardware deployment solutions.
  • Expanded control stack with MPC/VMC exploration and led Nav2 navigation framework integration.

Additional Awards

  • 2025 China Robot Competition & RoboCup China Open (FIRA Small Size): National Third Prize.
  • 27th China Robot and AI Competition: Sichuan First Prize and National Second Prize.

Research

Embodied Benchmark Dataset Practice (Second Author, ECCV 2026 under submission)

  • Built high-fidelity simulation and evaluation environments for sweeping and grasping tasks in Isaac Sim.

Multimodal Defect Detection with SFT+GRPO (Project Lead)

  • Developed a multimodal visual model for sewer defect detection.
  • Fine-tuned Qwen3-VL with optimized RL objectives combining IoU and format rewards.

Skills

  • Linux, Python, C/C++, ROS, MATLAB
  • PCB design, STM32/ESP32/ESP8266, Jetson Xavier deployment
  • SolidWorks, 3DBuilder, 3D printing

Video (Reserved)