STRUCTURE AND MECHANISM ESTIMATION OF AN ARTICULATED OBJECT BY MOTIONOBSERVATION

In this paper, we propose a method to construct structure models of ar ticulated objects from multiple local observations of their motion usi ng state transition analysis of local geometric constraints. The objec t model is constructed by a bottom-up approach with three levels. Each level groups sensor data with a constraint among local features obser ved by the sensor, and constructs the local model. If the sensor data in current model conflict, the model is reconstructed. In each level, the first level estimates a local geometric Feature from the local sen sor data (eg. edge, Feature point) The second level estimates a rigid body from the local geometric feature. The third level estimates an ob ject from the rigid bodies. In the third level, the constraint between rigid bodies is estimated by transition stales, which are motions bet ween rigid bodies. This approach is implemented on a blackboard system .