Agent-based Crowd Simulation Research Articles

Agent-based crowd simulation: an in-depth survey of determining factors for heterogeneous behavior

Agent-based crowd simulation: an in-depth survey of determining factors for heterogeneous behavior

Development of a Real-Time Crowd Flow Prediction and Visualization Platform for Crowd Management

Crowd management at large-scale events and specific facilities is a critical issue from the perspectives of safety and service quality improvement. Traditional methods for crowd management often rely on empirical knowledge, which has limitations in quickly grasping the on-site situation and making decisions on the spot. In this study, we developed a real-time crowd flow prediction and visualization platform incorporating an agent-based crowd simulation and an advanced crowd management system called crowd management platform as a service. In a case study focused on the area around the Tokyo Dome, we demonstrated that capturing pedestrian flow allows for accurate predictions of congestion at the nearest train station up to 10 min in advance. Moreover, the time required to predict the situation 20 min ahead for 3,000 agents was 1 min and 35 s, confirming the feasibility of real-time processing. To enhance the accuracy and reliability of the simulation results, a sensitivity analysis considering errors in pedestrian flow measurement revealed that simple linear models cannot capture the complexity of crowd dynamics adequately. Notably, the agent-based simulation replicated stop-and-go wave patterns observed in actual measurements under specific crowd conditions, confirming the advantage of using agent-based simulations. Finally, we proposed a method that enables facility managers and security personnel to conduct a more comprehensive evaluation. This method integrates their existing experience with the aggregated display of multiple simulation results, which includes consideration of errors in pedestrian flow measurement through a visualization platform.

BIM based framework for building evacuation using Bluetooth Low Energy and crowd simulation

Crowd evacuation is an area that has been studied by numerous researchers over the last years. Many models have been developed to simulate the evacuation. However, most of the developed models involved emergency routing of building occupants using hypothetical assumptions of their positions without integrating their actual locations into the models. Thus, in order to increase efficiency and accuracy of evacuation models, evacuees' real locations can be identified and then integrated into the model. In this research, Building Information Modelling (BIM), agent based simulation and Bluetooth Low Energy technology are integrated together to create an evacuation framework capable of indicating locations of building occupants and identifying the optimum evacuation path based on these locations. Point cloud data of a building is collected from 3D laser scanner, then converted to 3D BIM model, which is then imported into an agent based crowd simulation software. Also, an indoor positioning module is developed, which is composed of an indoor positioning system (IPS) developed by the authors. Through the IPS, evacuees’ locations are calculated, whereas Bluetooth emitters –beacons- held by evacuees emit signals that are detected by group of gateways mounted in the building detect. The locations are then sent to the crowd simulation software, then through integrating the modules, optimum evacuation route is identified, which is then converted to directions through a dynamic signage module. A case study is presented to illustrate the proposed framework. The results of the case study show that the developed indoor positioning system has an accuracy of 1.88 m with a maximum error range of 3.934 m. Moreover, the results indicate that the implemented indoor positioning system is capable of correctly detecting the evacuees in the rooms they are actually in with an accuracy of 93%. Finally, the simulation shows that the full evacuation of the floor under study took 2 min and 48 s.

Predictive Agent-Based Crowd Model Design Using Decentralized Control Systems

As a complex system, crowd dynamics emerge bottom-up from the local interactions between pedestrians as component subsystems. This article proposes a predictive agent-based crowd simulation model to analyze the outcomes of emergency evacuation scenarios taking into account collisions between pedestrians, smoke, fire sprinklers, and exit indicators. The crowd model is based on a decentralized control system structure, where each pedestrian agent is governed through a deliberative-reactive control architecture. The simulation model for evacuation includes a routing-based control system for dynamic-guided evacuation. A design case illustrates the modeling process. Results show that the crowd simulation model based on agent autonomy and local interactions is able to generate higher level crowd dynamics through emergence.

Crowd flow forecasting via agent-based simulations with sequential latent parameter estimation from aggregate observation

Unlike conventional crowd simulations for what-if analysis, agent-based crowd simulations for real-time applications are an emerging research topic and an important tool for better crowd managements in smart cities. Recent studies have attempted to incorporate the real-time crowd observations into crowd simulations for real-time crowd forecasting and management; however, crowd flow forecasting considering individual-level microscopic interactions, especially for large crowds, is still challenging. Here, we present a method that incorporates crowd observation data to forecast a large crowd flow, including thousands of individuals, using a microscopic agent-based model. By sequentially estimating both the crowd state and the latent parameter behind the crowd flows from the aggregate crowd density observation with the particle filter algorithm, the present method estimates and forecasts the large crowd flow using agent-based simulations that incorporate observation data. Numerical experiments, including a realistic evacuation scenario with 5000 individuals, demonstrated that the present method could successfully provide reasonable crowd flow forecasting for different crowd scenarios, even with limited information on crowd movements. These results support the feasibility of real-time crowd flow forecasting and subsequent crowd management, even for large but microscopic crowd problems.

Real-time agent-based crowd simulation with the Reversible Jump Unscented Kalman Filter

Commonly-used data assimilation methods are being adapted for use with agent-based models with the aim of allowing optimisation in response to new data in real-time. However, existing methods face difficulties working with categorical parameters, which are common in agent-based models. This paper presents a new method, the RJUKF, that combines the Unscented Kalman Filter (UKF) data assimilation algorithm with elements of the Reversible Jump (RJ) Markov chain Monte Carlo method. The proposed method is able to conduct data assimilation on both continuous and categorical parameters simultaneously. Compared to similar techniques for mixed state estimation, the RJUKF has the advantage of being efficient enough for online (i.e. real-time) application. The new method is demonstrated on the simulation of a crowd of people traversing a train station and is able to estimate both their current position (a continuous, Gaussian variable) and their chosen destination (a categorical parameter). This method makes a valuable contribution towards the use of agent-based models as tools for the management of crowds in busy places such as public transport hubs, shopping centres, or high streets.

SPH crowds: Agent-based crowd simulation up to extreme densities using fluid dynamics

In highly dense crowds of humans, collisions between people occur often. It is common to simulate such a crowd as one fluid-like entity (macroscopic), and not as a set of individuals (microscopic, agent-based). Agent-based simulations are preferred for lower densities because they preserve the properties of individual people. However, their collision handling is too simplistic for extreme-density crowds. Therefore, neither paradigm is ideal for all possible densities.In this paper, we combine agent-based crowd simulation with Smoothed Particle Hydrodynamics (SPH), a particle-based method that is popular for fluid simulation. We integrate SPH into the crowd simulation loop by treating each agent as a fluid particle. The forces of SPH (for pressure and viscosity) then augment the usual navigation behavior and contact forces per agent. We extend the standard SPH model with a dynamic rest density per particle, which intuitively controls the crowd density that an agent is willing to accept. We also present a simple way to let agents blend between individual navigation and fluid-like interactions depending on the SPH density.Experiments show that SPH improves agent-based simulation in several ways: better stability at high densities, more intuitive control over the crowd density, and easier replication of wave-propagation effects. Also, density-based blending between collision avoidance and SPH improves the simulation of mixed-density scenarios. Our implementation can simulate tens of thousands of agents in real-time. As such, this work successfully prepares the agent-based paradigm for crowd simulation at all densities.

Simulation and Evaluation of Precaution Strategies for COVID-19 Pandemic: A Case Study

The world now is living in Corona Virus Disease (COVID-19) pandemic era, which forced almost all businesses and companies to lockdown, which as a result led to economic fluctuation and social issues. Some precautions strategies were introduced and applied by governments to allow businesses to reopen to the public. This paper aims to model an agent-based crowd simulation using unity game engine with C# programming language in order to explore the effectiveness of the current precautions to limit the spread of COVID-19. Two important models are adopted in this work, which are the SIR disease spreading model and the homogeneous mixing behavior model. Utilizing these models, several parameters are chosen to control the viral spread among the agents in five different realistic scenarios. After several simulations, it was determined that the number of agents in the scene and wearing masks have the highest impact on limiting the viral spread.

Synchronizing navigation algorithms for crowd simulation via topological strategies

We present a novel topology-driven method for enhancing the navigation behavior of agents in virtual environments and crowds. In agent-based crowd simulations, each agent combines multiple navigation algorithms for path planning, collision avoidance, and more. This may lead to undesired motion whenever the algorithms disagree on how an agent should pass an obstacle or another agent.In this paper, we argue that all navigation algorithms yield a strategy: a set of decisions to pass obstacles and agents along the left or right. We show how to extract such a strategy from a (global) path and from a (local) velocity. Next, we propose a general way for an agent to resolve conflicts between the strategies of its algorithms. For example, an agent may re-plan its global path when collision avoidance suggests a detour. As such, we bridge conceptual gaps between algorithms, and we synchronize their results in a fundamentally new way. Experiments with an example implementation show that our strategy concept can improve the behavior of agents while preserving real-time performance. It can be applied to many agent-based simulations, regardless of their specific navigation algorithms. The concept is also suitable for explicitly sending agents in particular directions, e.g. to simulate signage.

A data-driven path planning model for crowd capacity analysis

In this paper, an agent-based crowd simulation model that focuses on path planning layer of (1) origin/destination popularities and (2) route choice is developed. This path planning model improves on the existing mathematical modeling and pattern recognition approaches by utilizing different sources of data to drive and validate it: video data was used for the open space scenarios and virtual reality experiments were applied for constrained space scenarios. For open space scenarios with video coverage, the density map of the video is extracted to calibrate the origin/destination popularities and the route probabilities among them. Factors related to space syntax, such as the traveling distance and turning angle, are proven effective features of the path planning model in this scenario. For constrained space scenarios, where the coverage of videos is usually limited, virtual reality experiments can be applied to learn the route choice model parameters at a fine granularity, particularly considering the crowdedness of the surroundings besides the space syntax factors. The navigation behaviors of players under different configurations in the virtual reality experiments were retrieved to train the route choice models using Support Vector Machine (SVM) model. The trained route choice model then simulates the crowd motion more realistically under different densities. We demonstrate the usefulness of the data-driven path planning model for crowd capacity analysis of a building layout.

IFC-centric performance-based evaluation of building evacuations using fire dynamics simulation and agent-based modeling

Abstract Despite recent technological advancements in the architecture, engineering, and construction (AEC) industries, the number of fire-caused fatalities in residential and commercial buildings in the United States over the last decade has consistently been an order of magnitude higher than fatalities caused by all other types of natural disasters combined. In this research, EvacuSafe is developed as a tool for designers to allow for layout optimization of buildings and facilities in terms of their evacuation safety performance. Contributions of this research include the development and validation of two spatiotemporal risk indices to quantify the safety of the egress routes and individual building compartments. Other unique advantages of EvacuSafe include the seamless integration of fire dynamics simulation, agent-based crowd simulation and building information models (BIM) using IFC data structures, allowing EvacuSafe to be readily used by designers to analyze a building layout design under various fire scenarios and for layout design optimization based on multiple safety criteria. The development of the framework, the risk indices, and the IFC-based integration is illustrated and validated through a case study on a two-story office building with 59 compartments. The results of the implementation show that the EvacuSafe is a valuable tool for evacuation design and planning that provides a more comprehensive evaluation of the evacuation performance in comparison to the existing indices and safety measures in the industry.

A Virtual Reality and Questionnaire Approach to Gathering Real-World Data for Agent-Based Crowd Simulation Models

Abstract Gathering real-world data is a crucial process in developing realistic, agent-based crowd simulation models. In order to gather real-world data, three types of data need to be considered: physical, mental, and visual. Existing data gathering methods do not collect all three data types, but they provide a limited amount of data for agent-based simulations. This article proposes using a combination of Virtual Reality and Questionnaires as a means to gathering real-world data. This hybrid method collects all three data types and is validated by comparing it to data collected from the real world. Two data gathering experiments (real world and our proposed method) were conducted to collect all three types of data for comparison. Experimental results show that the proposed method can collect similar data to the real-world experiment, in particular for mental and visual data. The Chi-Square Goodness-of-Fit Test proves that there is no significant difference between the real world and our proposed method for mental and visual data, whilst the test shows there is significant difference in physical data, in particular, completed time. We propose an adjustment factor for the completed time data that mitigates the gap between virtual space and real space, and allows the results collected to be input into agent-based simulations as real-world data. Overall, the proposed method is cost effective, time efficient, reproducible, ecologically valid, and able to collect three types of data for agent-based crowd simulation models.

Virtual reality crowd simulation: effects of agent density on user experience and behaviour

Agent-based crowd simulations are used for modelling building and space usage, allowing designers to explore hypothetical real-world scenarios, including extraordinary events such as evacuations. Existing work which engages virtual reality (VR) as a platform for crowd simulations has been primarily focussed on the validation of simulation models through observation; the use of interactions such as gaze to enhance a sense of immersion; or studies of proxemics. In this work, we extend previous studies of proxemics and examine the effects of varying crowd density on user experience and behaviour. We have created a simulation in which participants walk freely and perform a routine manual task, whilst interacting with agents controlled by a typical social force simulation model. We examine and report the effects of crowd density on both affective state and behaviour. Our results show a significant increase in negative affect with density, measured using a self-report scale. We further show significant differences in some aspects of user behaviours, using video analysis, and discuss how our results relate to VR simulation design for mixed human–agent scenarios.

Towards modeling pedestrian’s invisible trail for simulating crowd movement

Crowd modeling and simulation have drawn much attention in recent decades due to the functionality of recurrence of the crowd movement pattern in an efficient way. Much effort has been paid aiming at generating an accurate simulation result with respect to different aspects of crowd movement pattern. A fact has been observed that footprint left in mud or turf significantly affects pedestrian’s decision making and moving trajectory since those footprints help other pedestrians walk comfortably. Inspired by this, we in this paper propose a crowd simulation model aiming to model how the movement of previous pedestrians affects decision making process of the pedestrians coming later. Unlike pedestrians leaving footprint in mud or turf, pedestrians leave no marks on hard surface. We consider each step of pedestrian moving on hard surface as a mutable invisible footprint which further forms a virtual trail. We first build a model to simulate how the invisible footprint forms and evolves on hard surface, upon which an agent-based crowd simulation model is then built to simulate how pedestrian makes trade-off between the invisible trajectory and the shortest path. The proposed model is validated by case studies with two scenarios. The simulation results indicate that we are able to simulate the impact on pedestrian’s decision making by the invisible footprint.

The Buzz Metric: A Graph-based Method for Quantifying Productive Congestion in Generative Space Planning for Architecture

This paper describes a novel simulation method for measuring the amount of buzz or productive congestion in an interior architectural space. This “buzz metric” first predicts the amount and distribution of congestion based on a simulation of walking paths and, then, computes the extent to which this congestion is distributed across the space. Unlike dynamic methods for simulating occupant behavior such as agent-based crowd simulation, our method is computed statically, and thus is fast enough to be used in an automated generative design (multiobjective optimization) workflow. After describing how the metric is calculated, the paper demonstrates how it can be used in an automated generative design process through a case study of the design of an exhibit hall.

Agent-Based Crowd Simulation of Daily Goods Traditional Markets

Agent-Based Crowd Simulation of Daily Goods Traditional Markets

Crowd-driven mid-scale layout design

We propose a novel approach for designing mid-scale layouts by optimizing with respect to human crowd properties. Given an input layout domain such as the boundary of a shopping mall, our approach synthesizes the paths and sites by optimizing three metrics that measure crowd flow properties: mobility, accessibility, and coziness. While these metrics are straightforward to evaluate by a full agent-based crowd simulation, optimizing a layout usually requires hundreds of evaluations, which would require a long time to compute even using the latest crowd simulation techniques. To overcome this challenge, we propose a novel data-driven approach where nonlinear regressors are trained to capture the relationship between the agent-based metrics, and the geometrical and topological features of a layout. We demonstrate that by using the trained regressors, our approach can synthesize crowd-aware layouts and improve existing layouts with better crowd flow properties.

Learning behavior patterns from video for agent-based crowd modeling and simulation

This paper proposes a novel data-driven modeling framework to construct agent-based crowd model based on real-world video data. The constructed crowd model can generate crowd behaviors that match those observed in the video and can be used to predict trajectories of pedestrians in the same scenario. In the proposed framework, a dual-layer architecture is proposed to model crowd behaviors. The bottom layer models the microscopic collision avoidance behaviors, while the top layer models the macroscopic crowd behaviors such as the goal selection patterns and the path navigation patterns. An automatic learning algorithm is proposed to learn behavior patterns from video data. The learned behavior patterns are then integrated into the dual-layer architecture to generate realistic crowd behaviors. To validate its effectiveness, the proposed framework is applied to two different real world scenarios. The simulation results demonstrate that the proposed framework can generate crowd behaviors similar to those observed in the videos in terms of crowd density distribution. In addition, the proposed framework can also offer promising performance on predicting the trajectories of pedestrians.

Use of Agent-Based Crowd Simulation to Investigate the Performance of Large-Scale Intermodal Facilities: Case Study of Union Station in Toronto, Ontario, Canada

When planning complex transit terminals, hubs, and stations, it is critical to analyze a facility’s capacity to handle expected passenger movements and volumes. In Toronto, Ontario, Canada, the revitalization of Union Station, the country’s busiest transit facility, involved the development of a set of high-fidelity pedestrian microsimulation models that were used to plan the improvement of this major intermodal hub. The pedestrian models were first created with MASSMOTION software, and construction plans and projected transit schedules were applied to represent various scenarios during and after the station revitalization. These models were then calibrated and validated against data from passenger counts and transit usage surveys to provide an accurate base for future scenarios. Through the use of a fully constructed model of the station with 2031 transit demand projections, a series of stress experiments was performed to evaluate the station’s capability to handle future passenger volumes. Under projected 2031 circumstances, it was found that Union Station could operate reasonably well. However, an increase in inbound transit passenger volumes of a mere 10% over 2031 projections could lead to severe levels of congestion within constrained sections of the station. This paper reports on the development of the microsimulation models and the model results for future scenarios.

A motion planning method for simulating a virtual crowd

A model of motion planning for agent-based crowd simulation is one of the key techniques for simulating how an agent selects its velocity to move towards a given goal in each simulation time step. If there is no on-coming collision with other agents or obstacles around, the agent moves towards the designated goal directly with the desired speed and direction. However, the desired velocity may lead the agent to collide with other agents or obstacles, especially in a crowded scenario. In this case, the agent needs to adjust its velocity to avoid potential collisions, which is the main issue that a motion planning model needs to consider. This paper proposes a method for modelling how an agent conducts motion planning to generate velocity for agent-based crowd simulation, including collision detection, valid velocity set determination, velocity sampling, and velocity evaluation. In addition, the proposed method allows the agent to really collide with other agents. Hence, a rule-based model is applied to simulate how the agent makes a response and recovers from the collision. Simulation results from the case study indicate that the proposed motion planning method can be adapted to different what-if simulation scenarios and to different types of pedestrians. The performance of the model has been proven to be efficient.