| 摘要 |
<p>PROBLEM TO BE SOLVED: To provide a means for estimating a control signal corresponding to a state quantity detected from sensor information from the relationship between the state quantity accumulated by learning and the control signal.SOLUTION: A control signal Y corresponding to a state quantity X detected from sensor information is recorded. In a learning phase of an autonomous learning type controller, the state quantity detected from sensor information and corresponding set of control signals are sequentially recorded. When the state quantity is detected from the sensor information in an execution phase, a set of state quantities which are closest to the state quantity detected from the sensor information are selected and at the same time a set of control signals corresponding to the state quantity are also selected. Closeness (distance R) between the state quantity detected from the sensor information and the selected state quantity is calculated, and the control signal corresponding to the detected state quantity is estimated from the control signal corresponding to the closeness (distance) information and the selected state quantity.</p> |
