I've observed that in robotics space-partitioning is much more common to represent maps. I guess it's because with primitive partitioning you are basically doing a lot of precomputation with known primitives (you already know where everything is) whereas in robotics you add new primitives over time depending on where you are in space. Can ray-tracing techniques be used for mapping in applications like SLAM?
keenan
@hubbahubba There are a lot of subtle trade-offs between different spatial data structures; it could just be historical that folks in robotics use space partitioning. I'd be interested to know. I know that there are incremental schemes to build a BVH but that's slightly different than building one online. Good question.
I've observed that in robotics space-partitioning is much more common to represent maps. I guess it's because with primitive partitioning you are basically doing a lot of precomputation with known primitives (you already know where everything is) whereas in robotics you add new primitives over time depending on where you are in space. Can ray-tracing techniques be used for mapping in applications like SLAM?
@hubbahubba There are a lot of subtle trade-offs between different spatial data structures; it could just be historical that folks in robotics use space partitioning. I'd be interested to know. I know that there are incremental schemes to build a BVH but that's slightly different than building one online. Good question.