I am a PhD candidate at the Visual Computing Group, University of Konstanz, Germany, working on efficient visual SLAM, collaborative multi-agent SLAM, and embedded spatial perception for robotics and XR.

Over the past 15+ years, my work has spanned robotics perception, visual SLAM, sensor fusion, camera calibration, GPU-accelerated vision, real-time 3D systems, computer graphics, and XR. I focus on perception systems that connect geometry, graphics, efficient computation, and real-world deployment.

​

• Developing efficient visual SLAM methods for embedded and resource-constrained platforms, including hybrid indirect-direct tracking and GPU-accelerated photometric alignment.

​​​

• Exploring learning-based 3D reconstruction methods, including neural rendering and diffusion-based approaches, as complements to classical geometric SLAM pipelines.

 

• Implemented from scratch a Kalman-filter-based Bayesian Multi-Object Tracking (MOT) module for a safety-relevant ADAS LiDAR perception pipeline, including real-time tracking, data association, integration, and testing under operational constraints. The work was later extended toward radar-based and defense-related perception applications.

 

• Built high-fidelity VR-based simulation and training systems for defense applications, combining game-engine development with custom real-time 3D graphics applications for simulation, visualization, and debugging.

 

• Developed applied deep-learning methods for industrial image processing, including multi-spectral CNN-based classification of LCD glass surface defects. Recipient of the FokusFinder Best Thesis Award.

 

• Co-invented and developed a patented pinch/grab interaction method for depth-sensing systems, later widely adopted across VR, AR, and smart glasses.