How OpenUSD is smoothing the sim-to-reality gap for physical AI
One of the hardest problems with physical AI is that there are a wide range of ways to capture information about the world in ways that can be processed by computers. Thus, there are over eighty competing 3D formats today representing everything from physical structures, visual aspects, and the relations between things in the world. This creates an integration bottleneck when trying to make sense of data from cameras, lidar, schematics, CAD drawings, sensors, and other kinds of information. One promising standard has been OpenUSD stewarded under the auspices of the Alliance for OpenUSD (AFO), which released the first OpenUSD Core specification last December.
AFO has also spun up various interest and working groups to support workflows and use cases across various domains like materials, geometry, and physics. (‘Materials’ in this context means how light reflects off of things rather than chemistry – that one always throws me for a...
Copyright of this story solely belongs to diginomica.com. To see the full text click HERE
