Identification Of Lane-changing Intention And Study On Graded Lane-changing In Exit Diversion Zone Of Undersea Tunnel

Changing lanes is one of the most complex and dangerous driving operations in tunnels,so it is generally prohibited to change lanes in tunnels.With the increasing number of motor vehicles,urban tunnel traffic presents a congestion situation.Prohibiting vehicle lane changing in tunnels can reduce traffic efficiency,so it is important to study vehicle lane changing behavior in tunnels that not only ensures traffic safety but also improves traffic efficiency.In order to give full play to the functions of urban tunnels in the road network,tunnels usually have multiple exits,and vehicles are allowed to change lanes in the exit diversion area.Studying lane-changing intention recognition in tunnels and graded lane-changing in exit diversion zones has important research value for predicting driving behavior,reducing lane-changing risks and conflicts,and ensuring lane-changing safety.It also provides decision-making basis for advanced intelligent driving assistance systems.This study summarizes the research status of driver eye movement characteristics,lane-changing intention recognition,and exit diversion zone lane-changing at home and abroad.Firstly,real vehicle tests were used to collect vehicle operation data and driver eye movement data for undersea tunnels.Vehicle motion parameters such as vehicle speed data before and after lane-changing,turn signal on status,vehicle lateral offset angle,and eye movement parameters such as total gaze time,hot spot map,number of times of backtracking,and pupil diameter were selected for analysis.The vehicle motion laws and driver eye movement laws for three different lane changing types,namely,free left lane changing,free right lane changing,and forced right lane changing,were obtained.The types of undersea tunnel lane-changing are divided,and the lane-changing process is divided into three stages:environmental stimulation,lane-changing preparation,and lane-changing execution.Then,the plane grid of the driver’s field of view is divided,and the driver’s eye movement tracking path is extracted during the lane-changing preparation phase using the sliding spatiotemporal cuboid algorithm.The dynamic time warping center of gravity averaging algorithm(DBA)is used to optimize the eye tracking path.Using the K nearest neighbor algorithm(KNN),a lane-changing intention recognition model based on driver’s eye movement characteristics was constructed based on the dynamic temporal regularization distance of eye movement tracking paths.The results show that using DBA-KNN for lane-changing recognition has high accuracy,with the recognition accuracy of 87.5%,88.2%,and 90.1%for free left lane-changing,free right lane-changing,and forced right lane-changing,respectively.Finally,taking the Jiaozhou Bay undersea tunnel as an example,a graded lane-changing study was conducted on the outlet diversion zone of the undersea tunnel.The length and location parameters of transition sections,transition sections,and auxiliary lanes in the lane-changing area have been designed to form six different graded lane-changing schemes.The traffic environment scene of Jiaozhou Bay undersea tunnel was built using UC-Win Road software.Four evaluation indicators,namely,lane-changing position,lane-changing rate,pupil diameter,and vehicle speed change,were selected and analyzed.The weight of the indicators was determined using analytic hierarchy process and entropy weight method(AHP-EWM),and an evaluation index system was established.A comprehensive evaluation model for grading and lane changing in the exit diversion zone based on set pair analysis and approximate ideal ranking method(SPA-TOPSIS)was constructed.The results show that the optimal scheme is Scheme E,that is,the design parameters of the scheme are 290m for transition section I,210m for transition section II,120m for transition section,and 140m for auxiliary lane.The comprehensive evaluation index is?_E=0.968.