World Exploration Company is dedicated to pushing the boundaries of spatial-temporal perception. By leveraging advanced generative world models and continuous flow matching, we deliver highly efficient 4D occupancy forecasting solutions for the future of autonomous systems.
A foundational framework for constructing high-resolution 3D semantic voxel grids through multi-modal sensor fusion, providing precise static and dynamic environmental modeling.
Breaking the computational bottlenecks of traditional autoregressive methods. We utilize Continuous Flow Matching to generate physically coherent, long-horizon 4D occupancy predictions.
Integrating high-dimensional semantic voxel grids with low-level radar signals to guarantee absolute safety and high-fidelity environmental prediction under extreme weather conditions.
Supervised by Prof. Robert Laganière, Zeping specializes in the intersection of generative AI and autonomous driving. His current research addresses critical computational bottlenecks in 4D occupancy forecasting.
With extensive experience in deep learning, PyTorch, and managing multi-GPU architectures (RTX 4090 clusters), his work bridges theoretical computer vision with robust machine learning engineering.