The map matching technology used in car navigation systems compensates for car location error on the road map by comparing car location data, obtained from dead reckoning using a sensor as a gyro, with the road data in the map database. Conventional map matching technology has, however, a disadvantage that when a car travels in a parking area whose inside lanes are not registered in the road database, the car location is incorrectly adjusted using surrounding road data, and this causes a bad influence when determining the car location after leaving the parking area. Our new method was developed to reduce such data misadjustment due to the use of surrounding road data and to improve the accuracy of location calculation after leaving the parking area by recognising whether the car is running inside the parking area or not. The parking area is assumed to be a covered one such as an underground space or a multifloor parking building, where the GPS measurement and running pattern are distinctive. The method reported here can eliminate the misadjustment of a car's position due to the use of surrounding road data and reduce the estimation error of position after leaving the parking area down to 40% or less compared with the conventional method.
Samenvatting