Andrea Gorrini

Towards an integrated approach to crowd analysis and crowd synthesis

An adaptive model for the simulation of groups of pedestrians able to preserve their cohesion.An observation and analysis of groups of pedestrians in a real world situation.An integrated framework for crowd analysis and synthesis. Studies related to crowds of pedestrians, both those of theoretical nature and application oriented ones, have generally focused on either the analysis or the synthesis of the phenomena related to the interplay between individual pedestrians, each characterised by goals, preferences and potentially relevant relationships with others, and the environment in which they are situated. The cases in which these activities have been systematically integrated for a mutual benefit are still very few compared to the corpus of crowd related literature. This paper presents a case study of an integrated approach to the definition of an innovative model for pedestrian and crowd simulation (on the side of synthesis) that was actually motivated and supported by the analyses of empirical data acquired from both experimental settings and observations in real world scenarios. In particular, we will introduce a model for the adaptive behaviour of pedestrians that are also members of groups, that strive to maintain their cohesion even in difficult (e.g., high density) situations. The paper will show how the synthesis phase also provided inputs to the analysis of empirical data, in a virtuous circle.

Data Collection for Modeling and Simulation: Case Study at the University of Milan-Bicocca

The investigation of crowd dynamics is a complex field of study that involves different types of knowledge and skills, and, also from the socio-psychological perspective, the definition of crowd is still controversial. We propose to investigate analytically this phenomenon focusing on pedestrian dynamics in medium-high density situations, and, in particular, on proxemic behavior of walking groups. In this work we will present several results collected during the observation of the incoming pedestrian flows to an admission test at the University of Milano-Bicocca. In particular, we collected empirical data about: levels of density and of service, group spatial arrangement (degree of alignment and cohesion), group size and composition (gender), walking speed and lane formation. The statistical analysis of video footages of the event showed that a large majority of the incoming flow was composed of groups and that groups size significantly affects walking speed. Collected data will be used for an investigative modeling work aimed at simulating the observed crowd and pedestrian dynamics.

An Agent-based Model of Pedestrian Dynamics Considering Groups: A Real World Case Study

Groups have a significant influence of pedestrian dynamics: members of groups, in fact, carry out of a form of interaction (by means of verbal or non-verbal communication) that allows them to preserve cohesion even in particular conditions, such as counter flows, presence of obstacles or narrow passages. The paper describes an agent-based model for the simulation of crowds of pedestrians whose main innovative element is the representation and management of groups in the simulated population. The paper briefly describes the model and then it presents its application to a real world scenario in which an analysis of the impact of groups on the overall observed system dynamics was performed.

Empirical Investigation on Pedestrian Crowd Dynamics and Grouping

The definition and implementation of pedestrian simulation models requires empirical evidences, acquired by means of experiments and on-field observations, for sake of model calibration and validation. This paper describes an observation carried out in a urban commercial-touristic walkway (Vittorio Emanuele II Gallery, Milan, in collaboration with the Municipality of Milano). Although the analysis considered traditional metrics for describing pedestrian flow, such as the level of service, the main aim of this work was to quantify and characterize the presence, impact and behavior of groups in the observed population. In particular, we had confirmatory results on the frequency of groups in the observed situation, but we also achieved innovative results on trajectories and walking speeds: the walking path of individuals was 4 % longer than the average path of groups, but the average walking speed of group members was 37 % lower than the one of single pedestrians. Finally, a metric for characterizing group dispersion was defined and applied to the observed scenario: relatively large groups (size three and four) occupy more space in their movement when compared to couples. The achieved results represent useful empirical data for the calibration and validation of models for the simulation of pedestrians and crowd dynamics, but also for the development of automated techniques for data collection and analysis employing computer vision techniques.

An Innovative Scenario for Pedestrian Data Collection: The Observation of an Admission Test at the University of Milano-Bicocca

The investigation of crowd dynamics is a complex field of study that involves different types of knowledge and skills. Due to the difficulty in reaching an exhaustive definition of the notion of crowd, we propose to analytically investigate this phenomenon focusing on pedestrian dynamics in medium-high density situations, and, in particular, on proxemic behavior of walking groups. In this work we will present several results collected during the observation of the incoming pedestrian flows to an admission test at the University of Milano-Bicocca. In particular, we collected empirical data about: level of density and service, spatial arrangement, composition (size, gender) and walking speed of groups. The analysis of video footages of the event showed that unexpectedly a large majority of the incoming flow was composed of groups, and that group size significantly affects walking speed.

Age and Group-driven Pedestrian Behaviour: from Observations to Simulations

The development of pedestrian simulation systems requires the acquisition of empirical evidences about human behaviour for sake of model validation. In this framework, the paper presents the results of an on field observation of pedestrian behaviour in an urban crowded walkway. The research was aimed at testing the potentially combined effect of ageing and grouping on speed and proxemic behaviour. In particular, we focused on dyads, as the most frequent type of groups in the observed scenario. Results showed that in situation of irregular flows elderly pedestrians walked the 40% slower than adults, due to locomotion skill decline. Dyads walked the 30% slower than singles, due to the need to maintain spatial cohesion to communicate (proxemics). Results contributed to refine the parametric validation of the agent-based simulation system ELIAS38.

Group Dynamics in Pedestrian Crowds

Recent crowd disasters highlight the importance of properly planning and designing large urban events and public spaces to enhance the safety of people in a crowd during evacuations. Pedestrian crowd dynamics are empirically investigated with an interdisciplinary approach (i.e., social sciences, computer science, and traffic engineering) focusing on the effect of groups and their proxemic behavior and interactions while walking. Empirical evidence achieved from urban in-field observations and laboratory experiments are presented and compared. Results indicate that the proxemic behavior of walking groups has a negative effect on walking speed when flow is irregular (primarily because of the need for members to maintain spatial cohesion during locomotion). These results have important implications for the design of common metrics to characterize spatial interactions among pedestrians and for the validation of models to replicate crowd dynamics that consider the effects of groups under normal and emergency conditions.

Granulometric Distribution and Crowds of Groups: Focusing on Dyads.

Pedestrian flows are characterised by the preponderant presence of groups, with particular reference to dyads. The paper presents a research focused on group and age-driven pedestrian behaviour in an urban crowded scenario. Data analysis was performed by using an open source tracker tool. Results showed that in situation of irregular flows (LOS B) dyads walked 30 % slower than singles, and that elderly walked 40 % slower than adults. The achieved results have been used towards the validation of the simulation platform ELIAS 38, with reference to the representation of the granulometric distribution of groups and heterogeneous speed profiles.

Experimental investigation of pedestrian personal space

New generations of computer-based simulations of pedestrian crowd dynamics are aimed at improving both scientific knowledge about such complex social phenomena and practical use of the results for more efficient, safer management of pedestrian circulation dynamics in urban scenarios. Within this twofold framework, the validation of the developed models with field data plays a central role, and the acquisition of empirical evidence about human behavior during locomotion is mandatory. An empirical investigation of pedestrian personal space (i.e., the area surrounding an individual while walking and into which strangers cannot intrude without causing discomfort) is reported. The investigation was developed in an ad hoc experimental setting to collect meaningful data (e.g., measurements of the front zone of personal space under static and moving conditions). The main results are presented to illustrate both the adopted experimental setting and the related data-collection method. These results represent an innovative contribution toward defining a metric for the characterization of spatial interactions among pedestrians and allowing the validation of simulation models dedicated to the study of the dynamic behavior of pedestrian crowd phenomena.

Crossing Disciplinary Borders Through Studying Walkability

Computer-based simulations of pedestrian dynamics are aimed at improving the walkability of urban crowded scenarios, considering the pedestrians’ comfort and safety. The validation of the developed models requires a cross-disciplinary approach, and the acquisition of empirical evidences about human behavior is mandatory. The main purpose of this work is to report two case studies which allowed to perform simulations and validate the ELIAS38 agent-based computational model: (i) the naturalistic observation of pedestrian dynamics in an urban commercial-touristic walkway, focused on the impact of grouping and ageing on speed; (ii) the controlled experiment of pedestrian spatial behavior, focused on the impact of speed and cultural differences on personal space.