Karthik Ramani is trying to save engineers' time by reusing their knowledge in a 3D CAD shape-search database.
Present position: University Faculty Scholar, Professor of Mechanical Engineering, Purdue University
Degrees: B.Tech in mechanical engineering, Indian Institute of Technology (Madras, India); M.S. in mechanical engineering, The Ohio State University; Ph.D. in mechanical engineering, design division, Stanford University
How you describe your work at cocktail parties: The real world is full of information that needs to be searched with words. The missing link is searching with shapes. We're attempting to bridge the gap between what's in your head and what actually exists.
How does the shape-search system work?
We take a 3D model of a part and break it up into a bunch of small cubes called voxels, or volume elements. The voxels are then converted into a 'skeletal graph,' which has numbers that represent a part's shape. This hierarchical representation is searchable, so it allows a user to actually narrow-down shape distinctions. Within a few queries engineers can find the shape that they are after. Users can either draw a part to search for or they can choose from existing ones in the inventory.
How does it save time?
It's not like Google in that it can't be done in one shot. It's a multi-step process, but it works better than single-step systems. When we can access 3D models that already exist, we are reusing knowledge. Plus, another time benefit comes from not forcing the user to store information in folders, like in PLM (product lifecycle management) software.
What program requirements does it have?
It works independent of CAD systems, so it doesn't matter what program the drawing is in.
Is there a need for such a system?
Studies say engineers spend a lot of time looking for information. So something that gives engineers the opportunity to search for shapes is better than not having anything.