- upper/lower bound computation
- new nodes spawn
- compute point clouds' stats, normalize (For scalability)
-
set_rotation
for flattened k-d tree, to avoid repeated rotation computation -
branch
-feature/goicp-core/texture
: move data into texture objects to improve performance -
usecudaStreamSynchronize
instead ofcudaDeviceSynchronize
- use
float4
for Points, store squared sum as well
- ICP not converging