To let a robot device easily learn an action in a more realistic way.
This robot device is provided with a leaning part 103 learning a time-series signals obtained in an action part 106 according to an action when operating the action part 106 for exhibiting the action of a leg part, etc., a prediction part 104 predicting based on initial signals obtained in the action part 106 by an external force to the action part 106 after the learning, an action control part 102 comparing the prediction value of the prediction part 104 with the time-series signals learned by the learning part and controlling the action part 106 based on the comparison result. This constitution allows the robot device to remember the action by an input of the external force after the learning and exhibit the action.
JP2003126562 | BALLOON ROBOT |
WO/2016/067474 | ROBOT TOY |
JP2016021996 | STUFFED ANIMAL ROBOT |
SABE KOTARO
COSTA GABRIEL
OHASHI TAKESHI
Next Patent: SEMICONDUCTOR WAFER ID READER AND TRANSFER ROBOT WITH ROTATING FINGER