Search
On March 24th, a collaborative research team from the Hong Kong University of Science and Technology, Xiaomi Auto, and Huazhong University of Science and Technology introduced the Uni-Gaussians framework, a pioneering approach that facilitates the unified representation and divide-and-conquer rendering of Gaussian primitives in dynamic driving environments. This architecture leverages a dynamic Gaussian scene graph to model both static backgrounds and dynamic entities, encompassing vehicles and pedestrians. By rasterizing image data, it ensures high frame rate output, while LiDAR data employs Gaussian ray tracing to meticulously simulate laser pulse propagation. This innovation marks a significant advancement in the quality and computational efficiency of camera and LiDAR data simulation for autonomous driving scenarios.
-
C114 Communication Network -
Communication Home
