Infoscience

Thesis

Elaboration d'un système de navigation auto-alimenté (SNA) et évaluation de ses performances

This research deals with self-supplied navigation systems (SNS), a combination of the technology of gathering travel time using floating car data (FCD) and dynamic route guidance (DRG). Using similar on-board equipment, these two systems can easily be combined in the same vehicle. The main goal is to prove the relationship between the percentage of SNS-equipped vehicles and the performance that can be expected from the use of this technology. After the description of the SNS architecture, an evaluation method based on microscopic traffic simulation results is proposed. The different components of a SNS are introduced, particularly the statistics and treatment module, which estimates and predicts travel times from the data transmitted by equipped vehicles. An innovative approach, resulting from the disaggregated observation of these travel times, is proposed in order to improve existing estimation techniques. The description of the parameters influencing this estimation's performance is followed by an analysis of their combined impact. It stresses the necessity of adopting a combination of these parameters, depending on the equipment rate of the vehicles, in order to maximise the estimation's precision. Before the comparison between the performances of equipped and non-equipped vehicles, based on the Lausanne city centre road network, the realism of existing traffic assignment models is analysed. As a consequence, a new alternative for traffic assignment is proposed, consisting of an iterative approach based on the use of historic knowledge of a "typical" day and on a differentiation of three driver categories: standard, expert and tourist. The SNS performance evaluation, mainly in terms of travel time, shows that an equipment rate of only 1 to 2 0/00 is sufficient in order for equipped vehicles to show similar performances to standard drivers, this category representing the majority of drivers. An equipment rate of 5 to 50 0/00 is needed in order to pass above the expert category, which has a perfect knowledge of the road network. For higher equipment rates the benefit compared with the other drivers is less noticeable, but the overall performance of all vehicles is highly improved. Finally, a behaviour study of SNS-equipped vehicles in the case of an incident on the network shows certain limits implied by the fact that the guided vehicles and the ones providing traffic data are the same ones.

Related material