Robot operation based on a model updated by sensory data

Presents a robot manipulation system based on an environment model, in which the model is automatically adjusted to match the real world by means of sensory data. The sensor system has a hierarchical structure of two layers; a vision system which acquires the global workspace configuration, and a force-sensing system which measures precise position by contact detection using force-sensing. Combining these two sensors allows accurate position detection covering a wide range. The processes of measuring the real world, updating the environment model and resuming task operation are executed automatically during the running of job.<
>