Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Manufacturing
Siemens used machine learning to determine the precise assembly coordinates for inserting components into motor drives, building a database of successful positions to improve the robot's insertion accuracy despite product variations.
“So what we've done is used machine learning and to find out what is the assembly point to the coordinates for the robot. We are recording each time the robot is successful, their successful position, and then using that to improve the next insertion into the drive.”
Designcomputer algorithms
University of Bath researchers used computer algorithms to generate and optimize building shapes, determining the most efficient placement of beams and columns to minimize embodied carbon and material usage in concrete structures.
“We use computer algorithms to generate the shapes of buildings, and we use computer algorithms to then analyze them and assess whether they're really good shapes or bad shapes at being a building, and we use optimization techniques for that.”
Manufacturingcomputer algorithms
The Acorn project used computer algorithms to calculate the optimal, non-rectangular shapes for concrete components, allowing robots to fabricate these complex shapes efficiently in a factory setting to reduce material waste and embodied carbon.
“so acorn is, is using computers to work out what shape these things really need to be, and only put the material where it's needed for its structural performance. … robots don't care that they're crazy shaped.”