About me
I am a Master’s student in Mechatronics and Information Technology at Karlsruhe Institute of Technology (KIT), working at the intersection of embodied AI, robotics, and perception.
Research
My research interests lie in embodied AI and intelligent robotic systems.
I am interested in:
- Vision-Language Navigation (VLN) and Vision-Language-Action (VLA)
- Multimodal large language models (MLLMs) and agent-based systems
- Robot perception and 3D scene understanding
- Real-world robotic system development and deployment
Selected Projects
ROS2-Based Autonomous Exploration and Object Retrieval
Developed a ROS2-based robotic system for autonomous exploration and object retrieval on Boston Dynamics Spot. My primary contribution focused on the perception pipeline, including object detection and 3D localization.
- Trained and deployed a YOLOv5 model on a custom dataset (mAP@0.5 = 0.962)
- Built RGB-D perception modules for real-time object detection and 3D localization
- Implemented 2D-to-3D projection and TF-based coordinate transformation
- Designed multi-view fusion and temporal filtering
- Supported both global detection and close-range grasping
LLM-Based Heuristic Generation for Robotic Palletization
Developed an LLM-driven system for heuristic generation in robotic palletization (3D bin packing), integrating human-in-the-loop feedback and physical constraints.
- Designed an LLM interface for translating natural language into executable heuristics
- Built structured prompt templates incorporating constraints and runtime context
- Developed an API-based interaction pipeline
- Applied LoRA fine-tuning for efficient local model deployment
Technical Skills
Programming: Python, MATLAB, C++
Robotics: ROS2, TF, RViz, sensor integration
Perception: OpenCV, YOLO, RGB-D perception, multi-view fusion
Machine Learning: PyTorch, deep learning pipelines
Generative AI: LLM integration, prompt engineering, LoRA
Tools: Git, Linux, Docker