Skip to content
View fkyria's full-sized avatar
  • Cyprus
  • 17:14 (UTC +03:00)

Block or report fkyria

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
fkyria/README.md

About me:

Biorobotics engineer focused on robot perception and evaluation.
MSc research (University of Twente) explored using Vision-Language Models to autonomously evaluate robotic task execution; revealing how and where VLM attention breaks down during real-world manipulation tasks.

💻 What I Work With

Python PyTorch ROS2 HuggingFace Git MATLAB LaTeX

Projects

🔬 VLM-Based Robotic Task Evaluation

  • Autonomous evaluation framework using Qwen2.5-VL on a Franka Research 3 arm;
    found temporal attention as the core failure mode in long-horizon tasks

Connect with me

LinkedIn | Email

Pinned Loading

  1. qwen-vl-robot-task-evaluation qwen-vl-robot-task-evaluation Public

    Qwen-2.5 VL as an autonomous evaluator of robotic task execution using Dynamic Sequential Prompting.

    Python

  2. vlm-robotics-finetune vlm-robotics-finetune Public

    QLoRA fine-tuning of Qwen2-VL-2B on real robot manipulation VQA data (Robo2VLM-1).

    Jupyter Notebook

  3. second-opinion second-opinion Public

    Python