When Less is More: Avoiding Unnecessary Geometric Constraints in XR Task Instructions

When creating assembly instructions for 3D virtual objects in XR, users are often required to specify each object’s position and orientation (pose) relative to a parent coordinate system. However, there are some situations in which we might not care about the full pose of an object. And others in which only the relative relationship between a pair of objects might matter, rather than their relationships to other objects. Examples include providing instructions for another person or a robot system that has different constraints than we have and shouldn’t be restricted by our constraints. I will present an overview of approaches to address this that avoid specifying more information than is needed for a 3D task. And I will describe an XR user interface our lab is developing that supports mixing full specification of object poses for tasks that need it and only partial specification for tasks that don’t.