When a vehicle deviates from a guide route at a first branch point, the
state of the first branch point is memorized. Afterward, when a second
branch point becomes a target for a route guide and the state of the
second branch point is similar to the memorized state, a user of the
vehicle is provided with a guide for preventing deviation from the guide
route at the second branch point with a manner of the guide changed. This
procedure allows learning of tendency of the user with respect to a state
apt to cause the user to mistake a guide route. A guide to help prevent a
user from deviating from a guide route at a branch point can be thus
performed appropriately.