The Research Of Map Updating Based On Trajectory Data

Accurate road map plays a vital role in the correct navigation of vehicle.Existing road map cannot guarantee to be updated in real-time because of its high maintenance cost,and how to dynamically update the map according to the trajectory data becomes an important research problem.The existing approach is to infer the whole road map based on the trajectory data,which makes the inferred map information incomplete and inaccurate.Besides,it requires high computational overhead.There are also some ways to update new roads on existing maps,but the methods typically take a line-based or point-based strategy,and only consider that the trajectories are dense,which are also inefficient.Based on those map updating methods,a new method of new road discovery is proposed,which is a hybrid framework,called HyMU,and it combines the advantages of both line-based and point-based strategies.In addition,based on the data set of taxi trajectories,we conduct substantial quantitative evaluation and comparison experiments between HyMU and other similar methods.The experimental results show that HyMU is more efficient than other methods when processing sparse or low sampling rate trajectory data.Finally,we propose the problem of map updating in the trajectory sparse region,and our proposal combines the model of sliding window,which updates the inferred roads gradually on the map through continuous observation of the track in each time window.In addition,we conduct some experiments to evaluate the method of map updating.Finally,the main research work in this paper includes the following aspects:· Preprocessing of trajectory data Before data processing and mining,we need to remove some noise data,which can help us to process data more efficiently and improve the accuracy of final result.The data preprocessing in this paper includes two main phases,which are map matching,and denoising both in direction and distance.· A Hybrid new road discovery method There are often some roads that are not updated in a timely on the road map,and how to find those missing roads is very meaningful for navigating.The existing map updating methods discover new roads based on line-based or point-based strategy,which has advantages and shortcom-ings.This paper combines the advantages of line-segment and point based method,and proposes a hybrid new road discovery method.· Method of map updating in data sparse area A newly built road may not have too many vehicles driving on it,and some roads even have fewer trajectory data.As a result,how to deal with the data in sparse areas and be able to update the new roads found correctly on the map is also a work to be studied in this paper.