An autonomous vehicle becomes popular along the emergence of a self-driving car. However, the relation between a driver and the vehicle is not studied well. We are investigating the effect of an adaptation process on the particular relation with the vehicle. Although a machine learning technique can acquire the model of the driver, there are two design stances as the result of the adaptation; one is a machine as the part of the driver’s body. And the other is an autonomous machine which has an independent embodiment. We are revealing criteria to achieve one of the design stance by conducting an experiment.