Hello members of this community who have experience working with point clouds. 

I have a robot whose motions I am checking against a point cloud scan of its environment. What results in the better performance of the system in terms of computing power, speed and accuracy - using the point cloud as it is or going the next step and converting it to a mesh ? 

The reason why would also be greatly appreciated, if possible.

Robot moving towards point cloud, below

Robot in a state of collision with point cloud, below

Looking forward to your responses and discussions, 

Thanks in advance

Regards,

Omar