The prediction of a destination is of great use for advanced driver assistance systems (ADAS). However, clustering and predicting locations based only on latitude/longitude coordinates from cars lacks semantic information about locations and can be imprecise, especially once the car is parked. The functionality of these predictive systems can be greatly enhanced when the data used for predictions is extended beyond the vehicle and when predictions can already be made before entering the car. Therefore, we propose to include location-based services running on the driver’s smartphone and provide the prediction via a web service. The inclusion of explicit check-ins is used to extract semantic meaning of places and to improve predictions beyond the parking spot of the vehicle. We implemented a prototype, which combines data from foursquare and Google Latitude and enables predicting not only the locations but also the most probable times when a user will arrive at a certain place and when that user will leave this place. This information enables preconditioning and optimization of charging strategies for electric vehicles and can improve recommendation systems. This paper demonstrates the overall system architecture and explains the prototype implementation.
Pre- and Postdrive Predictions
Lect. Notes Electrical Eng.
2012-10-23
9 pages
Article/Chapter (Book)
Electronic Resource
English
Online Contents | 2011