Many times I've been driving at night and ran into street bumps at speed because they're not painted different or due to the lack of good light. In some cases this could harm your car. I thought of an application that might help avoid that.

Initially the app should contain 2 modes:

• Learn Mode
• Assist Mode

Learn Mode:

• Detect Bumps and record their location.
• Detect irregular parts/cracks in the street and record their location.
• For irregular areas, should ask whether it's only on the left, right, or the center of the streets wide.

Assist Mode

• Warn around approaching bump
• Warn about approaching irregular area, and on which side of the street.
• Warn about exceeding a pre-set speed

The app could be later extended to be a client for a full Web Service. Information recorded while learning could be stored on an online database, retrieved by users all around the world.

And don't worry about spam/fake info, since these are going to be auto corrected by other user's clients detecting those fake info. (Like expecting a street bump that never came, client app should immediately auto update the database with that info)

I still did not do any research on this, specially on whether information gathered from accelerometer could be used to identify bumps or cracks in the street (like matching it with a certain pattern or something).

I'd like to gather opinions of the community on this as I plan to propose it as project for GSOC 2010.