Real-time recognition of personal routes using instance-based learning

Abstract

Predicting routes is a critical enabler for many new location-based applications and services, such as warning drivers about congestion- or accident-risky areas. Hybrid vehicles can also utilize the route prediction for optimizing their charging and discharging phases. In this paper, a new lightweight route recognition approach using instance-based learning is introduced. In this approach, the current route is compared in real-time against the route instances observed in past, and the most similar route is selected. In order to assess the similarity between the routes, a similarity measure based on the longest common subsequence (LCSS) is employed, and an algorithm for incrementally evaluating the LCSS is introduced. The feasibility of the proposed approach is empirically evaluated using real-world data; the obtained results indicate that the routes can be accurately recognized with a delay of 11 turn-points.