Fill Value Passed to voxels_along_rays.

[darosaj]

I'm wondering if it may be easily possible to add a feature for passing in a fill value for returning the times from voxels_along_rays. So that if a ray doesn't intersect a voxel, it has a fill value that the user can define, rather than it returning the empty tensor.

I know this would be useful for the work flow of different "sensors" that interreacting with the voxel grid.

[fwilliams]

This seems a reasonable thing to add. Could you write out an example API with inputs and outputs of what you would like to have?

[darosaj]

I'm thinking something similar to this:

ray_o = JaggedTensor([ray_o])
ray_d = JaggedTensor([ray_d])

vox, times = grid.voxels_along_rays(ray_o, ray_d, dist_fill_value=1000, 1, 1e-4)

Where the fill value is passed in, and instead of returning empty tensors for the times (distances), if a fill value is provided, those tensors have that value.

Right now, I'm unbinding the times JaggedTensor, and iterating on that to fill the empty tensor, but would be much more efficent to have it filled before.

Thanks!