Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Interpretable geometry structures from sparse pointclouds
In this project we will be looking into efficient conversion of sparse pointcloud to representations that can be used directly for applications like path planning.
Sparse-SLAM methods often represent the scene as pointcloud and voxels. While this is efficient for tracking in SLAM, for downstream applications like
motion planning, this representation is often insufficient. Applications like motion
planning on the other hand prefer map representations like meshes or signed distance
function.In this project we will be looking into efficient conversion of sparse pointcloud to
meshes and signed distance function
Sparse-SLAM methods often represent the scene as pointcloud and voxels. While this is efficient for tracking in SLAM, for downstream applications like motion planning, this representation is often insufficient. Applications like motion planning on the other hand prefer map representations like meshes or signed distance function.In this project we will be looking into efficient conversion of sparse pointcloud to meshes and signed distance function
The goal is to efficiently generate high-level geometric structure, such as meshes and dense surface, from sparse voxel representation
The goal is to efficiently generate high-level geometric structure, such as meshes and dense surface, from sparse voxel representation