Automatic Procedure Generation

Problem Context

Humans or robots that have the task to build, repair, disassemble or maintain something benefit from clear work instructions that guide them through the process. Typical instructions comprise a series of steps that must be executed to reach a certain result, such as a manual to build a piece of furniture. Creating and updating these instructions requires expert knowledge, what can be a serious issue for companies that manage a lot of products.

Solution

3D CAD data contain geometrical and hierarchical information about the parts in a product. As such, it is often used a starting point to create (visual) instructions. However, information on how to build, repair or otherwise manipulate the product missing in these data. Traditionally, extensive design knowledge is needed to add this information. This project aims to infer this design knowledge from CAD and instruction examples with the use of AI. The developed algorithm will be used to make suggestions for step-by-step instructions.