| 摘要 |
Vehicle position information of a vehicle, image information of the surroundings of the vehicle, and feature information for the surroundings of the vehicle are obtained. An image recognition process is performed on a target feature contained in the image information. A behavior detecting unit detects a behavior of the vehicle that is performed within a predetermined range from a position in which the image recognition process has successfully been performed. Behavior detection information that indicates a result of the detection of the behavior is stored into a detection result storing unit, while being kept in correspondence with information of the detection position. Based on a plurality of pieces of behavior detection information that indicate mutually the same behavior of the vehicle and have been stored in the detection result storing unit while the vehicle is being driven through a same location a plurality of times, a learned behavior extracting unit extracts the repeatedly-detected behavior of the vehicle as a learned behavior and outputs, as learned behavior information, attribute information and position information of the learned behavior that are kept in correspondence with a piece of feature information of the corresponding target feature.
|
