Interaction Between Pedestrian And Autonomous At Unsignalized Mid-block Crosswalks Considering External Human-machine Interface

Pedestrians are among the most vulnerable and risk-taking road users at unsignalized mid-block crosswalks.The interaction between pedestrians and autonomous vehicles becomes a new challenge as high level autonomous vehicles no longer require a human driver.An external human-machine interface(e HMI)can effectively compensate for the lack of information.At the same time,the current overly conservative interaction strategy of autonomous vehicles may slow down road traffic.Based on the above,the purpose of this study is proposed: to explore safe and efficient interaction modes between pedestrians and autonomous vehicles unsignalized mid-block crosswalks,considering e HMI.Firstly,we designed 8 prototypes of e HMI with presentation information,and used E-prime to prepare an experimental program,classified participants into conservative and aggressive types by traffic behavior questionnaires.We recruited 24 participants to conduct picture stimulation experiments to investigate the influence of e HMI presentation information on pedestrian decision making,with the aim of verifying whether egocentric bias occurs in pedestrians and also finding acceptable allocentric message for e HMI;subsequently,we combined e HMI and vehicle kinematic cues to propose five different autonomous vehicles designs with different levels of conservatism,at the same time,eight interaction scenarios were designed unsignalized mid-block crosswalks,and 42 dynamic interaction scenarios were constructed using Unity 3D,and video experiments were conducted using E-prime to compile experimental programs to investigate the effect of autonomous vehicles with different levels of conservatism on decition of pedestrians,with the aim of verifying whether e HMI makes pedestrians focusist,and finding out the conservative design of autonomous vehicles that facilitates pedestrians to make rational decisions;finally,the interaction strategy between pedestrians and autonomous vehicles under unsignalized mid-block crosswalks is proposed.Using Net Logo simulation software,microscopic simulation experiments of pedestrians interacting with conservative or non-conservative autonomous vehicles are constructed on no zebra crossing section and zebra crosswalk sections,respectively.The impact of the degree of conservatism of autonomous vehicles on the capacity of unsignalized mid-block crosswalks is explored using vehicle and pedestrian delays as the primary evaluation metrics.The results show that 1)pedestrians mostly interpret e HMI from an egocentric perspective when faced with ambiguous information,i.e.,egocentric bias occurs.2)Vehicle intent is acceptable for e HMI information from a non-egocentric perspective.3)Pedestrians are more likely to obey e HMI at zebra crossing section compared to no zebra crossing section,i.e.,focalism occurs.4)At,autonomous vehicles adopting a non-conservative strategy can reduce vehicle delay no zebra crossing section time by 24.6% compared to the conservative strategy,while pedestrian waiting time is not significantly reduced.5)at zebra crossing section,the pedestrian waiting time is reduced by 24.48% and pedestrian throughput can be increased by 14.34% when autonomous vehicles adopt the non-conservative strategy,but it leads to an increase in vehicle delay time by 13.59%,but has little impact on the overall vehicle throughput.It is recommended that autonomous vehicle programmers adopt non-conservative and conservative strategies for no zebra crossing section and zebra crossing section,respectively,which can reduce the proportion of risky pedestrian crossings while improving interaction efficiency and guiding pedestrians to cross from crosswalk.