This paper presents a car navigation system able to plan a car's route and actual path. The route plan is built by integrating information acquired from an on-board road map knowledge base and information received from traffic monitoring centres. This information can be used to define a car's route through two different shortest path algorithms on the basis of optimization criteria (distance length, time length, touristic trip, and so on). When the criterion is the minimum time length, then the system takes advantage of a qualitative traffic flow model to evaluate the unknown traffic flows on the basis of the known traffic flows of the road network. Actual path plans can be built using two different planners. The first builds an accurate path taking into account both the external world and the car's movement constraints. The second builds a rough path which can be used both as an attempted path of the car, which will be refined by another planner, and for giving information to the driver. Nevertheless, while the first planner requires a large amount of sensor data and computation power, the second requires a limited set of data about road edges and obstacle distance and limited computation power. (A)
Abstract