Object Configurator

From TinyCog
Jump to: navigation, search

This subsystem is part of the 3D Reconstruction process. The object configurator contains 3D Scene Graph information (shape and texture) about objects (for example: a "Ford Mustang"), together with information about the internal state of the object (i.e. the specific model of the car, it's color, whether it's convertible and other parameters such as a damaged status etc). As a special case, the object configurator also contains 3D Scene Graph information about the facial expressions of a persons, together with the identified emotional states.

Main Operation

The main operation of the object configurator is to produce a 3D images of an Object, given the object's type and status information. This 3D image is used in the iterative recognition algorithm of the 3D Reconstruction subsystem as the reference for comparing with the actual sensor (camera) data.

In order to match sensor data with an internal model, the 3D Reconstruction algorithm will modify the object's internal parameters until an optimum match is found (or several non-optimum matches).

Comparison and References


Implementation Status

TinyCog does not yet implement an object configurator and there are no plans to do so within the scope of TinyCog. However, a different implementation of Scene Based Reasoning may focus on the interface with the real world and use this subsystem as a centerpiece.