| 摘要 |
The present invention makes it possible to inform a driver of unstable driving states even if learning of normal driving characteristics is not yet complete. On the basis of acquired driving-state data, a first unstable-driving determination unit in an information provision unit (100A) estimates a degree of driving instability from difference amounts between a plurality of driving-state distributions over different periods of time. A second unstable-driving determination unit in the information provision unit (100A) estimates a degree of driving instability using a process that is different from the estimation process used by the first unstable-driving determination unit. Using a learning level (SD) indicating the degree to which a driving-state distribution computed by a first driving-state-distribution computation means matches the driver's driving characteristics, a learning-completion determination unit in the information provision unit (100A) determines learning to be complete when a preset amount of learning time has passed from the start of the collection of the abovementioned driving-state data. When learning is complete, an instability-degree selection unit in the information provision unit (100A) selects the degree of instability estimated by the first unstable-driving determination unit, and when learning is not yet complete, said instability-degree selection unit selects the degree of instability estimated by the second unstable-driving determination unit. |
