Crowd Motion Research Articles

Unsupervised pattern and outlier detection for pedestrian trajectories using diffusion maps

The movement of pedestrian crowds is studied both for real-world applications and to gain fundamental scientific insights into systems of self-driven particles. Trajectory data describes the dynamics of pedestrian crowds at the level of individual movement paths. Analysing such data is a central challenge in pedestrian dynamics research, coupled with increasing data availability this implies a need for efficient methods to identify key features of the captured crowd dynamics. In this paper, we show that diffusion maps, an unsupervised manifold learning method, can be used for this purpose. We show how to build an informative feature space by defining a set of observables from trajectories. We use our diffusion map approach to analyse pedestrian movement on a stadium-shaped track, and during egress from a room, considering hundreds of trajectories for each scenario. We first verify that our diffusion map analysis can recover known leading variables that determine the system dynamics. Then, we show how our analysis facilitates a qualitative comparison of the dynamics inherent in entire data sets, by contrasting experimental with numerically simulated data. Finally, we establish how our approach can be used to automatically detect outliers that show behaviour distinct to others. These results indicate that our work can contribute a computationally efficient and unsupervised approach to analyse pedestrian dynamics without needing much prior knowledge of the data. We suggest this could be useful for automatically monitoring live data, or as an initial step to inform a subsequent analysis.

Illusory percepts of curvilinear self-motion when moving through crowds.

Self-motion generates optic flow, a pattern of expanding visual motion. Heading estimation from optic flow analysis is accurate in rigid environments, but it becomes challenging when other human walkers introduce independent motion to the scene. Previous studies showed that heading perception is surprisingly accurate when moving through a crowd of walkers but revealed strong heading biases when either articulation or translation of biological motion were presented in isolation. We hypothesized that these biases resulted from misperceiving the self-motion as curvilinear. Such errors might manifest as opposite biases depending on whether the observer perceived the crowd motion as indication of his/her self-translation or self-rotation. Our study investigated the link between heading biases and illusory path perception. Participants assessed heading and path perception while observing optic flow stimuli with varying walker movements. Self-motion perception was accurate during natural locomotion (articulation and translation), but significant heading biases occurred when walkers only articulated or translated. In this case, participants often reported a curved path of travel. Heading error and curvature pointed in opposite directions. On average, participants perceived the walker motion as evidence for viewpoint rotation leading to curvilinear path percepts.

Review Paper on IoT Thread Detection using Deep CNN Classifier

Abnormal activity will lead to uncommon changes in the crowd behavior. In other words, the crowd motion changes conform to certain rules for valid behaviors, while for abnormal events the motion changes are uncontrolled. The motion-changed rules to detect and localize abnormal behavior in crowd videos. Specifically, we first generate the motion patterns based on the descriptor of collectiveness. Then each frame pair is represented as a transfer matrix whose elements are the difference of a set of motion patterns. Thereafter, the motion-changed rules are constructed in the transformation space using a bag-of-words approach. Finally, the proposed approach measures the similarity between motion-changed rules and the incoming video data to examine whether the actions are anomalous. The approach is tested on the UMN dataset and a challenging dataset of crowd videos taken from the railway station. The experimental results demonstrate the effectiveness of the proposed method for detection abnormal behavior

Comparative analysis of the performance of humanitarian logistic structures using agent-based simulation

The post-disaster humanitarian logistic structure (PD-HL) used in a disaster determines the performance of the relief distribution, which impacts human suffering and loss of lives. This paper develops an Agent-Based simulation model to explore the performance of three different PD-HL structures: Agency Centric Efforts (ACEs), Partially Integrated Efforts (PIEs), and Collaborative Aid Networks (CANs). Using the proposed simulation framework, the authors analyze the role of information sharing and crowd movement in the distribution of relief supplies, and evaluate the ability of different PD-HL structures to respond to disasters of varying intensities. The performance of each PD-HL structure is measured using the social costs, i.e., the sum of logistic costs incurred by the relief organizations carrying out logistical activities, and the deprivation costs and walking costs of affected people. The study also discusses the humanitarian operations during the 2010 Port-au-Prince earthquake to demonstrate the real-world application of the simulation framework. The results suggest that sharing information can reduce the deprivation costs in general. However, as the level of information sharing increases, its marginal effect on deprivation costs decreases. Setting up temporary shelters to limit the movement of affected people can help reduce uncertainty in demands, thereby facilitating the distribution of relief supplies. The distribution capacity is the primary factor affecting the performance of the three PD-HL structures. The results indicate that CANs are particularly effective in responding to catastrophic disasters due to their high distribution capacity. However, ACEs and PIEs can also perform well in situations where their distribution capacities are adequate to ensure that available supplies reach the affected population. In addition, the waiting time of disaster-affected people at the points of distribution (PODs) significantly affects the distribution of relief supplies, while walking speed has no significant impact on the distribution of relief supplies.

Crowd model calibration at strategic, tactical, and operational levels: Full-spectrum sensitivity analyses show bottleneck parameters are most critical, followed by exit-choice-changing parameters

ABSTRACT Crowd motion simulation requires specification of a range of parameters, each reflecting certain aspects of agent behavior. But what parameters matter the most? Are they all equally important? The question is important given that available data and resources for parameter calibration are limited, and priorities often need to be made. Here, for the first time, a full-spectrum sensitivity analysis of crowd model parameters is reported. It is shown that estimates of simulated evacuation time are, by far, most dependent on the value of locomotion/operational parameters, especially those that determine discharge rate at bottlenecks. The next most critical set of parameters are those that influence change of direction choices. If a crowd simulation model fails to reproduce bottleneck flows accurately, efforts to refine other modeling layers will be in vain. Similarly, if the model fails to represent exit choice adaptation/changing accurately, efforts to refine the exit choice model will be fruitless.

Crowd Management Through Optimal Layout of Fences: An Ant Colony Approach Based on Crowd Simulation

The increasing population density in public places necessitates urgent attention to address safety concerns via effective crowd management. In many congested scenarios such as peak-hour subway stations, the utilization of fences to guide crowd movement has become a widely adopted approach to alleviate congestion. This work presents a method that combines crowd simulation and management, focusing on the optimization of the fence layout for efficient crowd guidance. First, a congestion probability social force model (CP-SFM) is introduced to simulate the irrational pedestrians and to evaluate the efficacy of different fence layouts. Second, based on CP-SFM, we are the first to formulate the fence layout problem as an optimization problem with the objective to minimize the congestion of pedestrians in public places. Third, we further propose an ant colony crowd intervention algorithm (ACCI) to optimize the layout of fences. Lastly, we illustrate the performance of proposed ACCI on 18 scenarios including two real-world subway stations. Compared with other optimization methods, ACCI demonstrates promising performance in avoiding crowd congestion.

Thermal–Acoustic Interaction Impacts on Crowd Behaviors in an Urban Park

As urbanization accelerates, parks, as vital urban public open spaces, and their acoustic and thermal ambience directly impact visitors’ comfort and the sustainability of parks. Selecting Xihu Park in Fuzhou, China located in the subtropical region as a typical example, this study utilizes covert observational experiments with different typical sounds (grass cutting, music, and no sound source) across temperature levels to examine the influence of thermal–acoustic interactions on crowd behaviors in the park. The findings are as follows: (1) melodious music can attract more tourists, while strong stimulating grass cutting noises under high temperatures reduce crowd flow. Excluding unpleasant audio sources, park soundscapes across temperatures have a relatively limited influence on attractiveness to people flow. (2) High temperatures diminish tourists’ interest in landscape experiences and persons staying, especially when the soundscape quality is poorer. Under non-high temperatures, audio environments have a minor impact on the staying time. (3) The soundscape quality plays a role by affecting people’s path choices of approaching or avoiding sound sources, where grass cutting noise has the most negative influence. Music, grass cutting sounds, and natural sounds demonstrate conspicuous differences in their effects under varied temperatures. (4) Comfortable acoustic environments can draw larger crowds and decrease the walking pace. High temperatures make crowds take slower steps. Different sound types have significant influences on crowd movement velocity under three typical temperature levels. This study comprehensively investigates the mechanisms of typical thermal–acoustic environments’ impacts on park crowd behaviors, providing important references for optimizing the acoustic and thermal environments of urban parks, while also enriching related research on landscape design and environmental psychology. Future studies can conduct in-depth explorations by creating more abundant thermal–acoustic combinations and probe differences across diverse populations.

Identifying critical urban intersections from a fine-grained spatio-temporal perspective

With the rising travel demand of crowds, urban road networks are under increasing pressure, which makes them progressively more fragile. Accurate identification of critical intersections is essential for urban management, resource allocation and traffic supervision. However, existing intersection evaluation methods fail to fully exploit transportation infrastructure, travel behavior and their synergistic relationships, resulting in coarse-grained and low-dynamic evaluation results. In this study, we develop an urban Intersection Evaluation framework from a Fine-grained Spatio-Temporal perspective (IE-FST). The framework is divided into three parts: fine-grained extraction of intersection information, multi-perspective assessment of node importance, and dynamic perception of spatio-temporal patterns in importance ranking. Specifically, IE-FST upgrades the intersection evaluation scale to turn-level, proposes the concept of Turning Sub-Node (TSN) and constructs a refined TSN topology network. Then, by finely extracting the network properties and mobility patterns of TSNs and mining the association patterns among TSNs, IE-FST proposes a Critical Fine-grained node Identification (CFI) algorithm to achieve a refined and dynamic evaluation of urban intersections. Finally, IE-FST analyzes the temporal characteristics and spatial distribution of TSN importance rankings, and provides traffic management measures for different types of TSNs to improve urban transport efficiency. Experiments are performed in Wuhan, China, and the results indicate that the proposed IE-FST framework achieves an accurate perception of urban intersections with high spatial and temporal resolution. This study promotes a better understanding of the synergistic relationships between crowd travel movements and urban transportation infrastructure, and provides theoretical support for infrastructure siting, refined traffic supervision, and rational resource allocation, thus contributing to the resilient and sustainable development of transportation systems and society.

Multi-Objective Deep Reinforcement Learning for Crowd Route Guidance Optimization

In this study, we propose an improved version of Pareto deep Q-network (PDQN), a multi-objective deep reinforcement learning method, and attempt to demonstrate its effectiveness in a real-world problem such as crowd route guidance strategy optimization. Overcrowding during crowd movement can sometimes lead to accidents; therefore, it is imperative to guide crowds to move safely and efficiently. Safety and efficiency are conflicting objectives, and how to dynamically determine guidance can be formulated as a multi-objective sequential decision-making problem. PDQN is suitable for solving these problems, but its applicability to complex real-world problems has not been fully verified. To apply PDQN to real problems, we propose to adjust the parameters of PDQN and improve the action selection criteria during learning. A toy problem and a crowd guidance problem using multi-agent crowd simulation are adopted to evaluate the performance of improved PDQN. Experimental results show that the improved PDQN can search for good strategies compared with the original PDQN and can obtain better strategies than simplified strategies.

Boosting urban community resilience to multi-hazard scenarios in open spaces: A virtual reality – serious game training prototype for heat wave protection and earthquake response

A key role in making cities resilient has been acknowledged in raising risk preparedness and awareness of urban communities, by appropriate education and communication strategies, which should rely on innovative and pervasive tools. In this regard, an outstanding paradigm shift is driven by the advancement of Virtual Reality, which can take advantage of Serious Games, for helping individuals develop responsive behaviours in case of both slow and sudden disasters and, thus, boosting effective human-urban-building interaction within a wider process of safety and sustainability. To this end, the paper proposes a VR-SG training prototype for multi-hazard scenarios in urban open spaces. The prototype integrates results from phenomenological and behavioural analyses and is applied to representative typologies of the built environment. The prototype is demonstrated for heat wave protection and earthquake response through the design and implementation of its functional features – virtual environment, interaction mode, learning outcomes and storyline – and its informative contents, including simulation-based data on surface temperatures, extent of falling debris and crowd motion. The final goal is to validate a reliable and flexible tool in view of wide replication in urban contexts for both instructing on critical situations and communicating mitigation strategies.

Effect of walking height on movement of individuals and crowds in a corridor

The safe and efficient evacuation of pedestrians in an emergency has been widely investigated in the past. Stooping is a common walking posture of pedestrians in evacuation. Understanding the basic laws of pedestrian movement with stoop posture is instructive for developing effective evacuation management in an emergency. However, the evacuation behaviors and movement characteristics of stooped pedestrians are still unclear. In this study, a series of experiments on individuals and crowds with different walking heights were conducted in an 18 m long corridor. It was found that a left-deflection phenomenon appears in the transition from upright walking to stooped walking, and the deflection angles of pedestrians decrease with increasing walking height. The influence of walking height on pedestrian velocity is revealed through the comparison of individual and crowd movement. Restricted walking height can promote the movement velocity of individuals at low speed levels, and weakens the velocity at high speed levels. In contrast to the individual, the velocities of crowds at different speed levels will be enhanced by restricted walking height. Generally, pedestrian movement is affected by the initial motivation, walking height and pedestrian density. Restricted walking height has an encouraging effect in the motivation of pedestrian movement but can increase the physical discomfort of pedestrians. Besides, crowd density has an inhibitory effect on pedestrian movement, and restricted walking height can alleviate high crowd density, implying that appropriate stoop movement can facilitate pedestrian traffic and evacuation to some extent.

Human Crowds as Social Networks: Collective Dynamics of Consensus and Polarization

A ubiquitous type of collective behavior and decision-making is the coordinated motion of bird flocks, fish schools, and human crowds. Collective decisions to move in the same direction, turn right or left, or split into subgroups arise in a self-organized fashion from local interactions between individuals without central plans or designated leaders. Strikingly similar phenomena of consensus (collective motion), clustering (subgroup formation), and bipolarization (splitting into extreme groups) are also observed in opinion formation. As we developed models of crowd dynamics and analyzed crowd networks, we found ourselves going down the same path as models of opinion dynamics in social networks. In this article, we draw out the parallels between human crowds and social networks. We show that models of crowd dynamics and opinion dynamics have a similar mathematical form and generate analogous phenomena in multiagent simulations. We suggest that they can be unified by a common collective dynamics, which may be extended to other psychological collectives. Models of collective dynamics thus offer a means to account for collective behavior and collective decisions without appealing to a priori mental structures.

Comparing Visual and Omniscient Models of Collective Crowd Motion

Comparing Visual and Omniscient Models of Collective Crowd Motion

An agent based modeling approach to evaluate crowd movement strategies and density at bathing areas during Kumbh Mela-2019

An agent based modeling approach to evaluate crowd movement strategies and density at bathing areas during Kumbh Mela-2019

Exploring single-line walking in immersive virtual reality

With increasing rates of elderly and obese people in the population, questions are being raised about the validity of inputs used by computer evacuation models to predict the movement of crowds in the built environment. The objective of this study is to examine the movement of individuals in a VR environment. Exploring individual movement in VR (where the individual is exposed to a virtual environment with virtual agents while actually moving alone in the physical environment) is a necessary step on the path to determining if VR is a useful tool to gather new crowd movement data. Specifically, this work presents the results of two experiments that were conducted to measure the correlation between inter-person distance (the distance from a participant to a virtual agent) and walking speed. Results show a positive correlation between walking speed and the inter-person distance for inter-person distances between 1.0 and 1.5 m. Above inter-person distances of 1.5 m, walking speed was not dependent on inter-person distance. An important finding from this work is no observed significant difference in the relationship between walking speed and inter-person distance across both experimental setups – ‘pushing’ or ‘following’ configurations. Finally, this work shows the potential of gathering individual movement data using VR.

Fish evacuate smoothly respecting a social bubble

Crowd movements are observed among different species and on different scales, from insects to mammals, as well as in non-cognitive systems, such as motile cells. When forced to escape through a narrow opening, most terrestrial animals behave like granular materials and clogging events decrease the efficiency of the evacuation. Here, we explore the evacuation behavior of macroscopic, aquatic agents, neon fish, and challenge their gregarious behavior by forcing the school through a constricted passage. Using a statistical analysis method developed for granular matter and applied to crowd evacuation, our results clearly show that, unlike crowds of people or herds of sheep, no clogging occurs at the bottleneck. The fish do not collide and wait for a minimum waiting time between two successive exits, while respecting a social distance. When the constriction becomes similar to or smaller than their social distance, the individual domains defined by this cognitive distance are deformed and fish density increases. We show that the current of escaping fish behaves like a set of deformable 2D-bubbles, their 2D domain, passing through a constriction. Schools of fish show that, by respecting social rules, a crowd of individuals can evacuate without clogging, even in an emergency situation.

Agent-Based Model of Mass Campus Shooting: Comparing Hiding and Moving of Civilians

The follow-up impact of the campus shooting incident is enormous. Due to a large number of minors (vulnerable groups) and complex building environments, the school is one of the main targets of terrorists. The Sandy Hook Elementary School shooting in 2012 poses a challenge to social and public security. Considering the dichotomous characteristics of the people shot and the impact of the refuge area on the crowd movement in the shooting incident, we use an agent-based model to restore the real target case. Based on the optimal solution, we explore the effects of the number and location of building exits, the scale of the shooter's attack, and the time to find shelter on the shooting results. For building exits, we found that for the environment in the case, the exit location has little impact on the number of casualties, but the increase in the number of exits provides more evacuation routes for the crowd and reduces the number of casualties. Panic will increase the probability of collision and injury. Hiding in a refuge area away from the shooter is effective to avoid shooting injury. At the same time, we propose the critical point of hiding probability. When it is greater than 40%, the faster the escape speed is, the higher the safety is; escape is more likely to replace avoidance in hiding probability less than 40% of environments. These findings are helpful to guide the building evacuation channel construction and emergency evacuation training.

Hajj pilgrimage abnormal crowd movement monitoring using optical flow and FCNN

This article discusses an effective technique for detecting abnormalities in Hajj crowd videos. In order to guarantee the identification of anomalies in scenes, a trained and supervised FCNN is turned into an FCNN using FCNNs and temporal data. By minimizing computational complexity, incorrect movement detection is utilized to achieve high performance in terms of speed and precision. This FCNN-based architecture is designed to handle two primary tasks: feature representation and the detection of incorrect movement outliers. Additionally, to overcome the aforementioned issues, this research will generate a new crowd anomaly video dataset based on the Hajj pilgrimage scenario. On the proposed dataset, the UCSD Ped2, Subway Entry, and Subway Exit datasets, the proposed FCNN-based technique obtained ultimate accuracy of 100%, 90%, 95%, and 89%, respectively. Additionally, the ResNet50-based technique achieved ultimate accuracy of 96%, 89%, 94%, and 92%, respectively, for the proposed dataset, the UCSD Ped2, Subway Entry, and Subway Exit datasets.

Analysis of the influence of detouring obstacle avoidance behavior on unidirectional flow

A pedestrian’s active obstacle avoidance behavior has an important impact on crowd safety in heterogeneous crowd movements, but how it affects the movement of individuals and crowd is not yet clear. This study proposes a detour anisotropic social force model (DASFM) that considers active obstacle avoidance behavior for future collision prediction. DASFM prioritizes moving obstacle pedestrians based on velocity obstacle (VO) and combines the anisotropic social force model (ASFM) to achieve the dynamic process of pedestrian active obstacle avoidance behavior. The results show that DASFM enables individuals to safely overtake while avoiding fluctuations in pedestrian speed and distance between pedestrians. The model improved abnormal “strip group” crowd aggregation, individual jostling, and congestion found in other models. It has effective performance in spatial distribution, trajectory, density evolution, and time consumption of crowd movement in unidirectional flow under open conditions. The fundamental diagram generated by DASFM in unidirectional flow under periodic boundary conditions tends to be consistent with the observations and experimental results of other researchers. Instantaneous flow analysis shows that overtaking behavior alleviates local congestion in the low-density crowd and promotes the slow flow of the high-density crowd. However, overtaking behavior also exacerbates the stop-and-go waves of the high-density crowd, posing a threat to crowd safety. This study not only effectively realizes the simulation of overtaking behavior among individuals and crowds, but also discovers the impact patterns of overtaking behavior on low- or high-density crowd. Therefore, measures should be taken to control individual overtaking behavior in high-density crowd to prevent the production of stop-and-go waves.

MSCDP: Multi-step crowd density predictor in indoor environment

Monitoring and predicting crowd movements in indoor environments are of great importance in crowd management to prevent crushing and trampling. Existing works mostly focused on individual trajectory forecasting in a less crowded scene, or crowd counting and density estimation. Only a very few works predict the crowd density distribution. However, this study is failing to realize multi-step prediction or exploits only density heatmaps modality and ignores the information complementation with corresponding video frames. Therefore, we are motivated to predict crowd density distribution in multiple time steps to facilitate long-term prediction. In this paper, a Multi-Step Crowd Density Predictor (MSCDP) to fuse video frame sequences and corresponding density heatmaps, is proposed to accurately forecast future crowd density heatmaps. To capture long-term periodic movement features, the long-term optical flow context memory (LOFCM) module is designed to store learnable patterns. We conducted extensive experiments on two real-world datasets. Evaluation results show that our MSCDP outperforms the state-of-the-art baseline techniques and MSCDP variants in terms of various prediction errors, demonstrating the effectiveness of MSCDP and each of its key components in multi-step crowd density prediction.