Portable Virtual Reality
US 7,113,618 uses motion inferred from a camera stream to update the viewpoint into a rendered virtual workspace. The key architectural move is to turn observed physical motion into stable control state for a virtual environment.
Augmented Reality System
US 7,301,547 extends the idea from motion estimation to scene understanding: visual observations are used to infer 3D information about the physical environment, construct a navigable representation, and place virtual content into that representation.
Architectural progression
2001: visual motion → viewpoint and control state.
2002: visual observations + inferred geometry → navigable world state.
What carried forward
The common systems problem is converting noisy sensor observations into a representation that downstream software can reason about efficiently. That thread later reappeared in camera and IMU interfaces, 3D scanning workflows, computer vision, and performance-sensitive sensor pipelines.