Reliable and accurate estimation of traffic states has significant social and scientific relevance. However, data for traffic estimation normally from different sources have different types, characteristics and so on, which makes it difficult to combine them into consistent and reliable traffic information. This thesis proposes efficient and parsimonious approaches called Data-Data Consistency to fuse heterogeneous traffic data. Results show that these approaches significantly improve the estimation accuracy and reliability. (Author/publisher)
Samenvatting