Motion forecasting is an essential task for autonomous driving, and utilizing information from infrastructure and other vehicles can enhance forecasting capabilities.

While these algorithms have been extensively studied, both theoretically and practically, see, e.g., [

While Diffusion-TT A is competitive, there's still room to unleash the full potential of diffusion models. To achieve this, two key aspects warrant further exploration.

We present a differentiable representation, DMesh, for general 3D triangular meshes. DMesh considers both the geometry and connectivity information of a mesh. In our design, we first get a set of convex tetrahedra that compactly tessellates the domain based on Weighted Delaunay Triangulation (WDT), and select triangular faces on the tetrahedra to define the final mesh.