| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a prediction device capable of predicting an occurrence frequency of a pedestrian or the like in a target area of prediction. <P>SOLUTION: After retrieving previous occurrence frequency data at S170, F<SB POS="POST">PFP</SB>is calculated at S180. Specifically, a prediction result of an output of a predictor is expressed as a vector. Prediction targets are classified into a total of twelve types which consist of a man, a woman, a child, a bicycle, an unknown object, and a dog at both the right and the left to set a feature vector (an objective variable of prediction) of a pedestrian occurrence frequency to twelve dimensions. When the feature vector is generated, status information becoming an explanatory variable of the vector consists of a location, a date, and weather. In the information, the location information consists of three dimensions in which each dimension is equivalent to a section, the date information consists of a year, a month, and a day of the week, which total twenty two dimensions, and the weather information consists of three dimensions of a sunny day, a cloudy day, and a rainy day. A regression relation of the feature vector to a status vector of the twenty eight dimensions is solved by a linear least square method. <P>COPYRIGHT: (C)2012,JPO&INPIT |
