Ongoing Research
Autonomous robot navigation and Agentic AI interaction in 3D space
Designed and developed a high-fidelity Unreal Engine simulation environment for autonomous agent training and evaluation. Created 3D space where multiple AI agents interact and navigate in a 3D environment
- Python ↔ Unreal Engine communication layer
- Collision-aware self-exploration without pre-training
- AlexNet-based perception model (~82% accuracy)
- *Research Paper will be published in March 2026*
Scene Flow for 3D characters
- Scene-flow pipeline streaming vision & skeleton data
- Capturing LOD of characters and streaming it to python code
- Creating pipeline to modify characters in Unreal Environment using python link
- Currently integrating metahuman plugin to modify characters
