Yufeng Jin

I am a PhD student at PEARL Lab at TU Darmstadt, jointly advised by Prof. Georgia Chalvatzaki and Dr. Mathias Franzius (Honda Research Institute Europe).
My research focuses on object 6D pose estimation and uncertainty modeling, with broader interests in robotics perception and human-object interaction.

Before starting my PhD in December 2023, I worked as a Robotics Perception Engineer at the Honda Research Institute Europe, where I conducted applied research in robot vision and teleoperation scenarios.

I received my Master’s degree in Mechatronics and Computer Science from the Karlsruhe Institute of Technology (KIT) in 2021.
My background bridges the domains of robotics and computer vision.

Email  /  Twitter  /  LinkedIn  /  Github

Research

An open-source agentic harness that turns a single natural-language prompt into ready-to-use reproduction, evaluation, fine-tuning, and deployment workflows for robot learning research — chambered, contract-typed, and validated by construction.

A probabilistic framework that leverages flow matching on the SE(3) manifold to estimate full 6D object pose distributions, enabling uncertainty-aware robotic manipulation under partial observability, occlusions, and symmetries.

A method that distills geometric features from pre-trained diffusion models via Manifold Distillation into a deterministic Spatial-Semantic Feature Pyramid Network, achieving geometrically consistent visuomotor control for robot manipulation with real-time performance.

A novel RL framework that explicitly leverages the inherent morphological symmetry in bimanual robotic systems to enable ambidextrous control.

A novel model-free framework for real-time 6D object pose estimation that leverages Gaussian Splatting for fast, accurate tracking and reconstruction from RGB-D input.

Study Notes · Paper Snapshots · learning notes & illustrated 中文 paper notes
Design and source code from Jon Barron's website.