有条件自动驾驶横向接管控制行为建模OA

[Background]In conditional automated driving,human drivers must take over when the system capability boundaries are reached.However,control instability and delayed responses often occur during the early takeover phase,compromising vehicle safety and efficiency of human-ma-chine coordination.[Objective]This study aims to characterize drivers'control recovery and envi-ronmental adaptation during takeovers,and to identify the key factors influencing takeover stabili-ty.[Method]A Carla-based driving simulation platform was developed to construct a fog-zone takeover scenario,and the information about participants'driving behavior,eye-tracking data,and subjective evaluations were collected.The sliding-window sample entropy(SW-SampEn)method was used to capture the control dynamics,from which two temporal indicators,takeover instability duration(TID)and environmental adaptation duration(EAD),were defined.A dual-Weibull accel-erated failure time(AFT)model incorporating clustered heterogeneity and gamma frailty was devel-oped to model the lateral takeover control behavior and stability characteristics of drivers.[Result]The results demonstrate that takeover control exhibits a"disruption-adaptation-stabilization"pat-tern of evolution.Synchronized traffic flow significantly prolongs both TID and EAD,while inter-rupted secondary tasks shorten the recovery time.Driving style,gender,experience,and trust in au-tomation significantly affect takeover stability.[Application]The proposed analytical framework provides methodological support and quantitative insights for takeover stability assessment,strate-gy design,and optimization of human-machine coordination,offering theoretical and practical refer-ences for safety management and control strategy development in complex automated driving envi-ronments.