Seekr
May - PresentCollaborative Research
Researcher & Frontend Developer
Seekr is a research prototype that explores how embodied AI agents perceive, remember, and navigate reconstructed 3D environments. Built with Gaussian Splatting and a vision-language model, the system allows an AI explorer to interpret human-guided visual observations, identify spatial relationships, and build an evolving representation of its surroundings.
<Tripo 3D, Java, Qwen, Gaussian Splatting>
<Tripo 3D, Java, Qwen, Gaussian Splatting>