Influence of A Cyclist on Pedestrian Movement on A Shared Road

共享道路上自行車影響下的行人運動實驗與模擬研究

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date6 Jul 2021

Abstract

Low-carbon transportation has been attracting wide attention from the public in recent years. Walking and cycling, as two important carbon-free modes of transportation, have become the main choices for people to travel over short distances. On the other hand, with the development of urbanization, urban roads are being broadened constantly, and the slow traffic lanes for pedestrian and cyclist are squeezed gradually. Thus the conflict between pedestrian and cyclist is becoming more and more serious. Therefore, it is necessary to carry out in-depth research on the pedestrian-cyclist mixed traffic on the shared road, so as to guarantee the comfort and safety for the travelling of pedestrian and cyclist with scientific facility design and traffic control. The thesis focuses on the macroscopic characteristics of pedestrian flow under the influence of a cyclist, the microscopic interaction between pedestrians and cyclists, and the feasibility of predicting pedestrian avoidance intention based on steps.

In the second chapter, we carried out a controlled experiment with a crowd of pedestrians and a cyclist involved on a circular track. The fundamental diagram, lane formation and formation of high density clusters of pedestrians with different densities were analyzed under the influence of the cyclist. The influence of crowd density and moving direction on the cyclist were investigated as well. The influence of the cyclist on the fundamental diagram of pedestrians shows evident difference under unidirectional and bidirectional scenarios. The cyclist restricts the pedestrian flow evidently in unidirectional flow. While in bidirectional flow, a critical transition density around 0.9ped/m2 is observed, where the restriction on the motion of crowd is not noticeable unless the critical density is reached. For cyclists, the speed also depends on the relative moving direction, but the discrepancy becomes marginal when the following effect is eliminated. The formed lanes are more concentrated in bidirectional flow than those in unidirectional flow. While the distances between the two lanes are larger in unidirectional flow. These differences can be explained by the cognition difference of pedestrians in the two scenarios. Besides, it was observed that a major effect of the cyclist on the crowd behavioral mechanism is the generation of backward propagated density wave. A Y-shaped pattern for the formation of high density clusters is discovered, which implied that stable high density clusters are likely to form as the forward propagated and backward propagated density waves merge. Moreover, in bidirectional flow, the generated backward propagated density wave could also promote the fading of stable high density clusters by reducing the heterogeneity of density distribution.

In the third chapter, a collision-free model for pedestrian movement in a corridor was developed and validated. Then the collision avoidance behaviors of pedestrian and cyclist in the interaction were studied through a series of experiments. In each run of the experiment, a group of single-file pedestrians and a cyclist moved unidirectionally or bidirectionally, Finally, combining with the discoveries in the experiment, we extended the collision-free model for pedestrian-cyclist interactions. The experiment indicated that the expected safety distance of cyclists is smaller than that of pedestrians. And there is a noticeable velocity change of pedestrians or cyclists during the collision avoidance process, which includes the path decision and the adjustment of speed. The behaviors depend on the distance between pedestrians and cyclists and their relative moving directions as well. Based on the experiment analysis, under the concept of subjective cognition and decision-making, we developed a collision-free model to describe the interactions among pedestrians and cyclists after introducing two parameters to decide the direction (critical headway distance and expected safety distance), and two parameters to decide the speed (relaxation time and critical deceleration distance). The respective lane formation characteristics of bidirectional and unidirectional pedestrian-cyclist mixed traffic are reproduced. The microscopic validation also implies that the proposed collision-free model can potentially be used to simulate the microscopic interaction between pedestrians and cyclists in bidirectional scenarios. The applicability of the model was not only at the trajectory level, but also at the level of instantaneous speed and moving direction.

In chapter 4, A data-driven approach was proposed for predicting the pedestrian avoidance intention based on steps. With a robot platform and a motion capture system, controlled experiments on the collision avoidance of pedestrians against a moving obstacle were conducted. The features related to avoidance steps were determined from the experimental results, and the feasibility of the data-driven approach was verified with the experimental data. Compared with the traditional modeling that only considers the headway distance, the data-driven approach can comprehensively deal with complex factors for better prediction. The AUC value indicates that with a reasonable feature vector selected, the classifier is capable to output plausible results. It is discovered that when a pedestrian intends to avoid the potential collision with an obstacle, it takes him two steps to adjust his movement. By including the state of the previous step into the feature vector, the performance of the approach is promoted. The proposed data-driven scheme demonstrates the applicability in other areas as well. The world have been suffering from the pandemic of Covid-19 when the study is being performed. Based on the proposed data-driven approach, we presented an intelligent test strategy under limited test capacity for the pandemic, and its superiority over the static strategy is demonstrated with the dataset in Lahore, Pakistan.

Finally, we summarized the study, discussed about the significance and limitations. Furthermore, the future work was planned.

    Research areas

  • Pedestrian dynamics, Mix traffic, Fundamental diagram, Lane formation, Pedestrian motion simulation, Behavior prediction