Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.668248
Title: An integrated solution based irregular driving detection
Author: Sun, Rui
ISNI:       0000 0004 5366 0707
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
Global Navigation Satellite Systems (GNSS) are used widely in the provision of Intelligent Transport System (ITS) services. Today, metre-level positioning accuracy, which is required for many applications including route guidance, fleet management and traffic control can be fulfilled by GNSS-based systems. Because of this level of success and potential, there is an increasing demand for GNSS to support applications with more stringent positioning requirements. These include safety related applications that require centimetre/decimetre level positioning accuracy, with high integrity, continuity and availability such as lane control, collision avoidance and intelligent speed assistance. Detecting lane level irregular driving behaviour is the basic requirement for lane level ITS applications. Currently, some research has addressed road level irregular driving detection, however very little research has been done in lane level irregular driving detection. The two major issues involved in the lane level irregular driving identification are access to high accuracy positioning and vehicle dynamic parameters, and extraction of erratic driving behaviour from this and the lane related information. This thesis proposes an integrated solution for the detection of lane level irregular driving behaviour. Access to high accuracy positioning is enabled by GPS and its integration with an Inertial Navigation System (INS) using Extended Kalman Filtering (EKF) and Particle Filtering (PF) with precise vehicle motion models and lane centre line information. Four motion models are used in this thesis: Constant Velocity (CV), Constant Acceleration (CA), Constant Turn Rate and Velocity (CTRV) and Constant Turn Rate and Acceleration (CTRA). The CV and CA models are used on straight lanes and the CTRV and CTRA models on curved lanes. Lane centre line information is extracted from defined lane coordinates in the simulation and is surveyed and stored as sets of positioning points from the motorway in the field test. The high accuracy vehicle positioning and dynamic parameters include yaw rate (omega) and lateral displacement (d) in addition to conventional navigation parameters such as position, velocity and acceleration. The detection of irregular driving behaviour is achieved by comparing the sorting rules of a driving classification indicator from the filter estimations with what is extracted from the reference. The detected irregular driving styles are characterized by weaving, swerving, jerky driving and normal driving on straight and curved lanes, based on the Fuzzy Inference System (FIS). The solution proposed in the thesis has been tested by simulation and validated by real field data. The simulation results show that different types of lane level irregular driving behaviour can be correctly identified by the algorithms developed in this thesis. This is confirmed by the application of data from a field test during which the dynamics of an instrumented vehicle supplied by Imperial College London were captured in real time. The results show that the precise positioning algorithms developed can improve the accuracy of GPS positioning and that the FIS based irregular driving detection algorithms can detect the different types of irregular driving. The evaluation of the designed integrated systems in the field test shows that a positioning accuracy of 0.5m (95%) source is required for lane level irregular driving detection, with a correct detection rate of 95% and availability of 94% based on a 1s output rate. This is useful for many safety related applications including lane departure warnings and collision avoidance.
Supervisor: Ochieng, Washington Yotto; Feng, Shaojun; Schuster, Wolfgang Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.668248  DOI: Not available
Share: