Article

Data-driven Crowd Modeling Techniques: A Survey

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Data-driven crowd modeling has now become a popular and effective approach for generating realistic crowd simulation and has been applied to a range of applications, such as anomaly detection and game design. In the past decades, a number of data-driven crowd modeling techniques have been proposed, providing many options for people to generate virtual crowd simulation. This article provides a comprehensive survey of these state-of-the-art data-driven modeling techniques. We first describe the commonly used datasets for crowd modeling. Then, we categorize and discuss the state-of-the-art data-driven crowd modeling methods. After that, data-driven crowd model validation techniques are discussed. Finally, six promising future research topics of data-driven crowd modeling are discussed.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Due to the simple structure and the clear physical meaning of the SFM and the corresponding parameters, modification and introduction of new forms of artificial forces can be easily made to improve the model's description ability of pedestrian movement process and phenomenon. Therefore, SFM is widely adopted in both literatures and popular pedestrian simulation platforms [19]. However, no matter how many modifications and enhancements haven been made to the evolution of the generic SFM, the mathematical expression of the driving force 0 ...
... Table 6, Table 9, and Table 11. Where, SI is defined as the standard deviation of ranks of performance of f across experiments of being different pedestrian density level.SI , as stated in Eq.(19), is a statistic for measuring the dispersion of performance rank for a certain assigning strategy of the desired speed across the ...
Article
Full-text available
In this paper, the desired speed is figured out as an important parameter for ordinary differential equation (ODE) based pedestrian simulation models. However, there lacks a thorough study on the optimal way of assigning the desired speed for ODE based pedestrian simulation models to achieve reliable performance. In order to gain a deep understanding of the role of the desired speed and its influence on the performance of the ODE based pedestrian simulation models, a total of nine assigning strategies of the desired speed are conducted a comparison study. Specifically, the performance of the ODE based models equipped with such desired speed assigning strategies are compared with regard to reference data from both the uni- and bi-directional pedestrian flow scenarios using a fair model comparison framework. Then, the performances are compared across several dimensions, including different assigning strategies of the desired speed, the different pedestrian simulation models, different experiments for the same motion base, and different motion base scenarios. Based on these results, four findings are summarized. Firstly, the desired speed is quantitively verified to be an important role in the performance of ODE based pedestrian simulation models. Secondly, pedestrian density level is verified to be an effective indicator for the complexity of the pedestrian crowd dynamics. Thirdly, the performance of the ODE based models is dependent on the extent to which the heterogeneity of desired speed is considered. Lastly, the performance of SF is suggested to be more predictable than ORCA regarding stability and optimality. Finally, based on the above findings, recommendations on assigning the desired speed for ODE based pedestrian simulation models are given.
... In the literature there have been several models [9] proposed for crowd simulation that can be grouped into major categories such as microscopic (rule-based, force-based, velocity-based, agent-based, vision-based) [10], macroscopic (continuum models, aggregate dynamics, potential eld-based models) as well as mesoscopic models (dynamic group behavior, interactive group formation, social psychological crowds). In terms of future challenges and directions such as generally increasing realism of simulations [11], apart from approaches and improvements such as data collection and integration, datadriven crowd modelling [12,10], cognitive-based crowds, learning to generate character animation [10], using machine learning approaches, computational eciency through parallel computing is of key importance in order to simulate large-scale realistic crowds. ...
Article
In the paper we present a modern efficient parallel OpenMP+CUDA implementation of crowd simulation for hybrid CPU+GPU systems and demonstrate its higher performance over CPU-only and GPU-only implementations for several problem sizes including 10 000, 50 000, 100 000, 500 000 and 1 000 000 agents. We show how performance varies for various tile sizes and what CPU–GPU load balancing settings shall be preferred for various domain sizes among CPUs and GPUs of a high performance system with 2 Intel Xeon Silver multicore CPUs and 8 NVIDIA Quadro RTX 5000 GPUs. We then present how execution time depends on the number of agents as well as the number of CUDA streams used for parallel execution of several CUDA kernels. We discuss the design and implementation of an algorithm with CPU computational threads, GPU management threads, assignment of particular tasks to threads as well as usage of pinned memory and CUDA shared memory for maximizing performance.
... Therefore, movement behavior of humans needs to be addressed accordingly. In some fields of application, e.g., pedestrian dynamics, movement research is extensive (Zhong et al. 2022, Zhou et al. 2010). However, we still lack a general framework for studying why, how, where, and when human agents move. ...
Article
Full-text available
Computer simulation of dense crowds is finding increased use in event planning, congestion prediction, and threat assessment. State-of-the-art particle-based crowd methods assume and aim for collision-free trajectories. That is an idealistic yet not overly realistic expectation, as near-collisions increase in dense and rushed settings compared with typically sparse pedestrian scenarios. Centroidal particle dynamics (CPD) is a method we defined that explicitly models the compressible personal space area surrounding each entity to inform its local pathing and collision-avoidance decisions. We illustrate how our proposed agent-based method for local dynamics can reproduce several key emergent dense crowd phenomena at the microscopic level with higher congruence to real trajectory data and with more visually convincing collision-avoidance paths than the existing state of the art. We present advanced models in which we consider distraction of the pedestrians in the crowd, flocking behavior, interaction with vehicles (ambulances, police) and other advanced models that show that emergent behavior in the simulated crowds is similar to the behavior observed in reality. We discuss how to increase confidence in CPD, potentially making it also suitable for use in safety-critical applications, including urban design, evacuation analysis, and crowd-safety planning.
Article
Full-text available
Data visualization is a key tool in data mining for understanding big datasets. Many visualization methods have been proposed, including the well-regarded state-of-the-art method t-distributed stochastic neighbor embedding. However, the most powerful visualization methods have a significant limitation: the manner in which they create their visualization from the original features of the dataset is completely opaque. Many domains require an understanding of the data in terms of the original features; there is hence a need for powerful visualization methods which use understandable models. In this article, we propose a genetic programming (GP) approach called GP-tSNE for evolving interpretable mappings from the dataset to high-quality visualizations. A multiobjective approach is designed that produces a variety of visualizations in a single run which gives different tradeoffs between visual quality and model complexity. Testing against baseline methods on a variety of datasets shows the clear potential of GP-tSNE to allow deeper insight into data than that provided by existing visualization methods. We further highlight the benefits of a multiobjective approach through an in-depth analysis of a candidate front, which shows how multiple models can be analyzed jointly to give increased insight into the dataset.
Article
Full-text available
This paper presents a comparison and relative assessment of data driven techniques and conventional theoretical pedestrian simulation models. Data driven methods are applied for simulating phenomena without a priori knowledge of parameter connections, while indicating high modeling performance. In contrast, theoretical models rely on pedestrian kinematics principles and provide mathematical functions on model parameters. A comparison between locally weighted regression and a social force model in real data (collected by the authors within the framework of this research) suggests superior performance of the data driven model on modeling pedestrian movements. However, a more integrated comparative analysis should be conducted, to validate these preliminary observations. Additional contributions, presented in this research, include an algorithm for eliminating data noise, based on an Unscented Kalman filter and moving average extensions.
Article
Full-text available
This paper analyzes the emergent behaviors of pedestrian groups that learn through the multiagent reinforcement learning model developed in our group. Five scenarios studied in the pedestrian model literature, and with different levels of complexity, were simulated in order to analyze the robustness and the scalability of the model. Firstly, a reduced group of agents must learn by interaction with the environment in each scenario. In this phase, each agent learns its own kinematic controller, that will drive it at a simulation time. Secondly, the number of simulated agents is increased, in each scenario where agents have previously learnt, to test the appearance of emergent macroscopic behaviors without additional learning. This strategy allows us to evaluate the robustness and the consistency and quality of the learned behaviors. For this purpose several tools from pedestrian dynamics, such as fundamental diagrams and density maps, are used. The results reveal that the developed model is capable of simulating human-like micro and macro pedestrian behaviors for the simulation scenarios studied, including those where the number of pedestrians has been scaled by one order of magnitude with respect to the situation learned.
Article
Full-text available
In moving pedestrian crowds, the distribution of individuals over different available routes emerges from the decisions of individuals that may be influenced by the actions of others. Understanding this phenomenon not only is important for research into collective behaviour, but also has practical applications for building safety and event management. Here, we study the mechanisms underlying pedestrian route choice, focusing on how time-independent information, such as path lengths, and time-dependent information, such as queue lengths, affect both initial decisions and subsequent changes in route choices. We address these questions using experiments with nearly 140 volunteers and an individual-based model for route choice. Crucially, we consider a wide range of route choice scenarios. We find that initial route choices of pedestrians achieve a balanced usage of available routes. Our model suggests that pedestrians performing trade-offs between exit widths and predicted exit crowdedness can explain this emergent distribution in many contexts. Few pedestrians adjust their route choice in our experiments. Simulations suggest that these decisions could be explained by pedestrians comparing estimates of the time it would take them to reach their target using different routes. Route choice is complex, but our findings suggest that conceptually simple behaviours may explain many movement decisions.
Article
Full-text available
As humans we possess an intuitive ability for navigation which we master through years of practice; however existing approaches to model this trait for diverse tasks including monitoring pedestrian flow and detecting abnormal events have been limited by using a variety of hand-crafted features. Recent research in the area of deep-learning has demonstrated the power of learning features directly from the data; and related research in recurrent neural networks has shown exemplary results in sequence-to-sequence problems such as neural machine translation and neural image caption generation. Motivated by these approaches, we propose a novel method to predict the future motion of a pedestrian given a short history of their, and their neighbours, past behaviour. The novelty of the proposed method is the combined attention model which utilises both "soft attention" as well as "hard-wired" attention in order to map the trajectory information from the local neighbourhood to the future positions of the pedestrian of interest. We illustrate how a simple approximation of attention weights (i.e hard-wired) can be merged together with soft attention weights in order to make our model applicable for challenging real world scenarios with hundreds of neighbours. The navigational capability of the proposed method is tested on two challenging publicly available surveillance databases where our model outperforms the current-state-of-the-art methods. Additionally, we illustrate how the proposed architecture can be directly applied for the task of abnormal event detection without handcrafting the features.
Article
Full-text available
We propose a new semantic-level crowd evaluation metric in this paper. Crowd simulation has been an active and important area for several decades. However, only recently has there been an increased focus on evaluating the fidelity of the results with respect to real-world situations. The focus to date has been on analyzing the properties of low-level features such as pedestrian trajectories, or global features such as crowd densities. We propose the first approach based on finding semantic information represented by latent Path Patterns in both real and simulated data in order to analyze and compare them. Unsupervised clustering by non-parametric Bayesian inference is used to learn the patterns, which themselves provide a rich visualization of the crowd behavior. To this end, we present a new Stochastic Variational Dual Hierarchical Dirichlet Process (SV-DHDP) model. The fidelity of the patterns is computed with respect to a reference, thus allowing the outputs of different algorithms to be compared with each other and/or with real data accordingly. Detailed evaluations and comparisons with existing metrics show that our method is a good alternative for comparing crowd data at a different level and also works with more types of data, holds fewer assumptions and is more robust to noise.
Article
Full-text available
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.
Article
Full-text available
We present Menge, a cross-platform, extensible, modular framework for simulating pedestrian movement in a crowd. Menge's architecture is inspired by an implicit decomposition of the problem of simulating crowds into component subproblems. These subproblems can typically be solved in many ways; different combinations of subproblem solutions yield crowd simulators with likewise varying properties. Menge creates abstractions for those subproblems and provides a plug-in architecture so that a novel simulator can be dynamically configured by connecting built-in and bespoke implementations of solutions to the various subproblems. Use of this type of framework could facilitate crowd simulation research, evaluation, and applications by reducing the cost of entering the domain, facilitating collaboration, and making comparisons between algorithms simpler. We show how the Menge framework is compatible with many prior models and algorithms used in crowd simulation and illustrate its flexibility via a varied set of scenarios and applications.
Article
Full-text available
Visual recognition of crowd dynamics has had a huge impact on several applications including surveillance, situation awareness, homeland security and intelligent environments. However, the state-of-the-art in crowd analysis has become diverse due to factors such as: a) the underlying definition of a crowd, b) the constituent elements of the crowd processing model, c) its application, hence d) the dataset and e) the evaluation criteria available for benchmarking. Although such diversity is healthy, the techniques for crowd modelling thus developed have failed to establish credibility therefore becoming unreliable and of questionable validity across different research disciplines. This paper aims to give an account of such issues by deducing key statistical evidence from the existing literature and providing recommendations towards focusing on the general aspects of techniques rather than any specific algorithm. The advances and trends in crowd analysis are drawn in the context of crowd modelling studies published in leading journals and conferences over the past 7 years. Finally, this paper shall also provide a qualitative and quantitative comparison of some existing methods using various publicly available crowd datasets to substantiate some of the theoretical claims.
Conference Paper
Full-text available
Crowd simulation has been an active and important area of research in the field of interactive 3D graphics for several decades. However, only recently has there been an increased focus on evaluating the fidelity of the results with respect to real-world situations. The focus to date has been on analyzing the properties of low-level features such as pedestrian trajectories, or global features such as crowd densities. We propose a new approach based on finding latent Path Patterns in both real and simulated data in order to analyze and compare them. Unsupervised clustering by non-parametric Bayesian inference is used to learn the patterns, which themselves provide a rich visualization of the crowd's behaviour. To this end, we present a new Stochastic Variational Dual Hierarchical Dirichlet Process (SV-DHDP) model. The fidelity of the patterns is then computed with respect to a reference, thus allowing the outputs of different algorithms to be compared with each other and/or with real data accordingly.
Technical Report
Full-text available
We present a novel dense crowd simulation method. In real crowds of high density, people manoeuvring the crowd need to twist their torso to pass between others. Our proposed method employs capsule-shaped agents, which enables us to plan such torso orientations. Contrary to other crowd simulation systems, which often focus on the movement of the entire crowd, our method distinguishes between active agents that try to manoeuvre through the crowd, and passive agents that have no incentive to move. We introduce the concept of a focus point to influence crowd agent orientation. Recorded data from real human crowds are used for validation, which shows that our proposed model produces equivalent paths for 85% of the validation set. Furthermore, we present a character animation technique that uses the results from our crowd model to generate torso-twisting and side-stepping characters.
Article
Full-text available
Many visual surveillance tasks, e.g. video summarisation, is conventionally accomplished through analysing imagery-based features. Relying solely on visual cues for public surveillance video understanding is unreliable, since visual observations obtained from public space CCTV video data are often not sufficiently trustworthy and events of interest can be subtle. We believe that non-visual data sources such as weather reports and traffic sensory signals can be exploited to complement visual data for video content analysis and summarisation. In this paper, we present a novel unsupervised framework to learn jointly from both visual and independently-drawn non-visual data sources for discovering meaningful latent structure of surveillance video data. In particular, we investigate ways to cope with discrepant dimension and representation whilst associating these heterogeneous data sources, and derive effective mechanism to tolerate with missing and incomplete data from different sources. We show that the proposed multi-source learning framework not only achieves better video content clustering than state-of-the-art methods, but also is capable of accurately inferring missing non-visual semantics from previously-unseen videos. In addition, a comprehensive user study is conducted to validate the quality of video summarisation generated using the proposed multi-source model.
Article
Full-text available
This paper proposes a crowd dynamic macroscopic model grounded on microscopic phenomenological observations which are upscaled by means of a formal mathematical procedure. The actual applicability of the model to real world problems is tested by considering the pedestrian traffic along footbridges, of interest for Structural and Transportation Engineering. The genuinely macroscopic quantitative description of the crowd flow directly matches the engineering need of bulk results. However, three issues beyond the sole modelling are of primary importance: the pedestrian inflow conditions, the numerical approximation of the equations for non trivial footbridge geometries, and the calibration of the free parameters of the model on the basis of in situ measurements currently available. These issues are discussed and a solution strategy is proposed.
Article
Full-text available
Automated scene analysis has been a topic of great interest in computer vision and cognitive science. Recently, with the growth of crowd phenomena in the real world, crowded scene analysis has attracted much attention. However, the visual occlusions and ambiguities in crowded scenes, as well as the complex behaviors and scene semantics, make the analysis a challenging task. In the past few years, an increasing number of works on crowded scene analysis have been reported, covering different aspects including crowd motion pattern learning, crowd behavior and activity analysis, and anomaly detection in crowds. This paper surveys the state-of-the-art techniques on this topic. We first provide the background knowledge and the available fea-tures related to crowded scenes. Then, existing models, popular algorithms, evaluation protocols, as well as system performance are provided corresponding to different aspects of crowded scene analysis. We also outline the available datasets for performance evaluation. Finally, some research problems and promising future directions are presented with discussions.
Article
The paper consists of 27 figures; numerous equations and 12 notes/ references, many of which are written by the authors of this paper. Whilst this may indicate a lack of “reading around the subject” it also indicates the unique nature of the topic and that little exists at present in the public domain about this topic. Indeed the authors and the research group they represent are the main contributors to the IMOs discussions and circulars on this subject. Given that background the paper is very detailed and consists of comparisons between the evacuation times of 3 passenger ships, 2 being Ro-Pax vessels and 1 a cruise liner. On board evacuation time statistics have been gathered from significant populations enabling the authors to draw significant conclusions relating to evacuation times in the presented scenarios. The paper is therefore a useful addition to the debates on this subject which is of major relevance to the understanding of evacuation times in passenger vessels. Data and research in this area is difficult to obtain thus the authors should be congratulated for their work.
Article
This paper provides a critical review on the state-of-the-art pedestrian and evacuation dynamics so as to comprehensively comprehend the motion behaviors of pedestrians from observations to simulation aspects. Types of typical data collection methods, namely the field survey, the controlled experiment, and the animal experiment, are classified, and the connections and differences of these three observation methods are explored. Pedestrians' complex behaviors characterized by the self-organization phenomena and movement data characterized by the fundamental diagram are then studied after the data collections, which can be used to calibrate and validate the pedestrian models. The mathematical models for pedestrian dynamics from both tactical level and operational level are also highlighted. The simulation data produced by the mathematical models could further reproduce pedestrian behaviors during the observations and contribute to decision makings for improving the evacuation efficiency. The applications of pedestrian models for behavior analysis, evacuation simulation, and layout design are also presented. Some challenges and future directions in the pedestrian and evacuation dynamics are also put forward. Findings presented in this study are helpful for researchers who want to understand the pedestrian and evacuation dynamics and to perform further research in this field.
Article
Simulation models are widely used to study complex systems. Current simulation models are generally handcrafted using expert knowledge (knowledge-driven); however, this process is slow and introduces modeler bias. This article presents an approach towards data-driven simulation modeling by developing a framework that discovers simulation models in an automated way for mobile agent-based applications. The framework is comprised of three components: (1) a model space specification, (2) a search method (genetic algorithm), and (3) framework measurement metrics. The model space specification provides a formal specification for the general model structure from which various models can be generated. The search method is used to efficiently search the model space for candidate models that exhibit desired behavior patterns. The five framework measurement metrics: flexibility, comprehensibility, controllability, composability, and robustness, are developed to evaluate the overall framework. The results demonstrate that it is possible to discover a variety of interesting models using the framework.
Conference Paper
Over the last two decades there has been a proliferation of methods for simulating crowds of humans. As the number of different methods and their complexity increases, it becomes increasingly unrealistic to expect researchers and users to keep up with all the possible options and trade-offs. We therefore see the need for tools that can facilitate both domain experts and non-expert users of crowd simulation in making high-level decisions about the best simulation methods to use in different scenarios. In this paper, we leverage trajectory data from human crowds and machine learning techniques to learn a manifold which captures representative local navigation scenarios that humans encounter in real life. We show the applicability of this manifold in crowd research, including analyzing trends in simulation accuracy, and creating automated systems to assist in choosing an appropriate simulation method for a given scenario.
Article
In this article, we propose a role-dependent (RD) data-driven modeling approach to simulate pedestrians’ motion in high-density scenes. It is commonly observed that pedestrians behave quite differently when walking in a dense crowd. Some people explore routes toward their destinations. Meanwhile, some people deliberately follow others, leading to lane formation. Based on these observations, two roles are included in the proposed model: leader and follower. The motion behaviors of leader and follower are modeled separately. Leaders’ behaviors are learned from real crowd motion data using state-action pairs, while followers’ behaviors are calculated based on specific targets that are obtained dynamically during the simulation. The proposed RD model is trained and applied to different real-world datasets to evaluate its generality and effectiveness. The simulation results demonstrate that the RD model is capable of simulating crowd behaviors in crowded scenes realistically and reproducing collective crowd behaviors such as lane formation.
Article
Extensive skill training using simulations for improving decision making has become a common practice in a multitude of domains and professions due to the expense or danger of training in real-life circumstances, or to speed up the learning process. In this work we propose a novel approach called PATDS (Performance Analysis of Training through Data-Farming Simulations) in which a simulated agent as a virtual copy of a human trainee is constructed and tested in all circumstances to provide feedback and decision support for another round of human learning. PATDS includes the analysis of behavior demonstrated by the trainee in simulations and the identification of relevant behavior patterns; that is, strategies. The identified patterns are in turn used to model a surrogate agent of the trainee. The created agent is then evaluated with data-farming experiments on a variety of simulation parameters to estimate the key weaknesses in its behavior and the overall performance. In subsequent rounds of training, the trainee improves the behavior based on the identified weaknesses and strategic decision recommendations; a new surrogate agent is designed adding new skills from the trainee until the desired overall performance is achieved. We evaluated the proposed approach on two asymmetric urban-terrain threat scenarios where a squad of security forces interacts with hostile crowds. We demonstrate that the data-farming-based evaluation in the complete simulation parameter space is beneficial for improving the trainees’ capabilities.
Article
Nowadays, there are many different types of NPCs (Non-Playable Character) in games, which can be endowed with different abilities, levels of intelligence and more or less realistic motion. How much the NPCs abilities are sophisticated, more complex are techniques to model and control them. Concerning the realism of NPCs motion, the ideal result is to provide characters that act and move in a way that seems natural to people. However, it can take a long time and a lot of work for developers, since designers should define characteristics for each different background character. This work aims to introduce a trajectory learning system derived from real video sequences, which can be used to generate NPCs motion for games. In addition to NPCs trajectories, we propose to detect in video sequences the formed groups and its characteristics to be used to simulate more realistic NPCs. We performed some evaluations using a pre-defined set of real video sequences and compared them with generated simulations. The obtained results indicate that our motion learning and groups detection techniques can be a valuable resource in order to generate NPCs and groups of NPCs trajectories, in an automatic way and similar to the real-life.
Article
As artificial intelligence becomes a research hotspot, more and more researchers are trying to apply it to numerous domains. It is therefore interesting and challenging to apply data-driven neural network technology to pedestrian movement modeling to test its effect against traditional social force model which can be applied to manifold pedestrian scenarios. Recent neural network based pedestrian movement simulation studies often train the network with only one scenario and then test within this scenario with various parameters. To make a more adaptive neural network, we propose a four layer network to learn multiple scenario data by normalization of relative positions among pedestrians, transferring velocity vector to scalar and incorporating more path planning information, and thus to make it more adaptive to various scenarios. Simulation results show that the proposed neural network approach can be applied to several typical pedestrian scenarios including counterflow and evacuation. Moreover, it shows more realistic speed-density curve and generates less trajectory fluctuations compared with social force model. Therefore, the proposed method is capable of generating more realistic pedestrian flow in multiple scenarios.
Article
This paper addresses the problem of efficiently simulating a believable virtual crowd. Our method is the first one that uses the Neural Network (NN) model to fit behaviors from real crowd data to a crowd simulation. Unlike several rule-based approaches that often result in ‘walking robots’, our model can learn motion rules derived from real data and later simulate human walking motions. Additionally, unlike the existing data-driven crowd simulation methods that have to perform search operations on the bound dataset simultaneously during the simulation, our model directly uses the NN model to generate the proper motion for each crowd member. The proposed method is being tested on various scenarios and compared with state-of-the-art state-action-based methods that are commonly employed in data-driven crowd simulation systems. The results demonstrate a significant increase in speed, as well as better simulation quality.
Article
This paper proposes a video data-driven social force model for simulating crowd evacuation. The initialization of pedestrian position, path navigation, and goal selection in the improved social force model was guided by real video data. To initialize pedestrian position and determine path navigation, the distribution of the pedestrians is set according to the real video. We also extracted the trajectories of pedestrian movement from the videos, and these trajectories were stored into a path set to guide the evacuation of pedestrians. Moreover, a fitness function was defined to model the behavior of a pedestrian goal selection. The fitness function could process the evacuation parameters, which were extracted from the video, and consider the degree and distance of exit congestion. Furthermore, we quantified the relationship values among pedestrians, and a new force called “group force” was added to the primary social force model. Pedestrians with close relationship gathered into one group and walked together. To validate the effectiveness of the proposed method, the video data-driven model was applied to simulate campus halls and roads. Simulation results show that the proposed approach is consistent with real-world situations and can assist in analyzing emergency evacuation scenarios.
Article
We propose a data-driven approach for tuning, validating and optimizing crowd simulations by learning parameters from real-life videos. We discuss the common traits of incidents and their video footages suitable for the learning step. We then demonstrate the learning process in three real-life incidents: a bombing attack, a panic situation on the subway and a Black Friday rush. We reanimate the incidents using an existing emotion contagion and crowd simulation framework and optimize the parameters that characterize agent behavior with respect to the data extracted from the video footages of the incidents.
Article
We propose an automatic crowd control framework based on multi-objective optimisation of strategy space using genetic programming. In particular, based on the sensed local crowd densities at different segments, our framework is capable of generating control strategies that guide the individuals on when and where to slow down for optimal overall crowd flow in realtime, quantitatively measured by multiple objectives such as shorter travel time and less congestion along the path. The resulting Pareto-front allows selection of resilient and efficient crowd control strategies in different situations. We first chose a benchmark scenario as used in [1] to test the proposed method. Results show that our method is capable of finding control strategies that are not only quantitatively measured better, but also well aligned with domain experts’ recommendations on effective crowd control such as “slower is faster” and “asymmetric control”. We further applied the proposed framework in actual event planning with approximately 400 participants navigating through a multi-story building. In comparison with the baseline crowd models that do no employ control strategies or just use some hard-coded rules, the proposed framework achieves a shorter travel time and a significantly lower (20%) congestion along critical segments of the path.
Article
Evacuation planning is an important and difficult task in building design. The proposed framework can identify optimal evacuation plans using decision points, which control the ratio of agents that select a particular route at a specific spatial location. The authors optimize these ratios to achieve the best evacuation based on a quantitatively validated metric for evacuation performance. This metric captures many of the important aspects of an evacuation: total evacuation time, average evacuation time, agent speed, and local agent density. The proposed approach was validated using a night club model that incorporates real data from an actual evacuation.
Article
In this paper, we present a data-driven approach to simulate realistic locomotion of virtual pedestrians. We focus on simulating low-level pedestrians' motion, where a pedestrian's motion is mainly affected by other pedestrians and static obstacles nearby, and the preferred velocities of agents (direction and speed) are obtained from higher level path planning models. Before the simulation, collision avoidance processes (i.e. examples) are extracted from videos to describe how pedestrians avoid collisions, which are then clustered using hierarchical clustering algorithm with a novel distance function to find similar patterns of pedestrians' collision avoidance behaviours. During the simulation, at each time step, the perceived state of each agent is classified into one cluster using a neural network trained before the simulation. A sequence of velocity vectors, representing the agent's future motion, is selected among the examples corresponding to the chosen cluster. The proposed CLUST model is trained and applied to different real-world datasets to evaluate its generality and effectiveness both qualitatively and quantitatively. The simulation results demonstrate that the proposed model can generate realistic crowd behaviours with comparable computational cost.
Article
Motivation: Integrating different omics profiles is a challenging task, which provides a comprehensive way to understand complex diseases in a multi-view manner. One key for such an integration is to extract intrinsic patterns in concordance with data structures, so as to discover consistent information across various data types even with noise pollution. Thus, we proposed a novel framework called "pattern fusion analysis" (PFA), which performs automated information alignment and bias correction, to fuse local sample-patterns (e.g., from each data type) into a global sample-pattern corresponding to phenotypes (e.g., across most data types). In particular, PFA can identify significant sample-patterns from different omics profiles by optimally adjusting the effects of each data type to the patterns, thereby alleviating the problems to process different platforms and different reliability levels of heterogeneous data. Results: To validate the effectiveness of our method, we first tested PFA on various synthetic datasets, and found that PFA can not only capture the intrinsic sample clustering structures from the multi-omics data in contrast to the state-of-the-art methods, such as iClusterPlus, SNF and moCluster, but also provide an automatic weight-scheme to measure the corresponding contributions by data types or even samples. In addition, the computational results show that PFA can reveal shared and complementary sample-patterns across data types with distinct signal-to-noise ratios in Cancer Cell Line Encyclopedia (CCLE) data sets, and outperforms over other works at identifying clinically distinct cancer subtypes in The Cancer Genome Atlas (TCGA) data sets. Availability: PFA has been implemented as a Matlab package, which is available at http://www.sysbio.ac.cn/cb/chenlab/images/PFApackage_0.1.rar . Contact: lnchen@sibs.ac.cn. Supplementary information: Supplementary data are available at Bioinformatics online.
Article
Designing suitable behavioral rules of agents so as to generate realistic behaviors is a fundamental and challenging task in many forms of computational modeling. This paper proposes a novel methodology to automatically generate a descriptive model, in the form of behavioral rules, from video data of human crowds. In the proposed methodology, the problem of modeling crowd behaviors is formulated as a symbolic regression problem and the self-learning gene expression programming is utilized to solve the problem and automatically obtain behavioral rules that match data. To evaluate its effectiveness, we apply the proposed method to generate a model from a video dataset in Switzerland and then test the generality of the model by validating against video data from the United States. The results demonstrate that, based on the observed movement of people in one scenario, the proposed methodology can automatically construct a general model capable of describing the crowd dynamics of another scenario in a different context (e.g., Switzerland vs. U.S.) as long as that the crowd behavior patterns are similar.
Article
Crowd video analysis has applications in crowd management, public space design, and visual surveillance. Example tasks potentially aided by automated analysis include anomaly detection (such as a person walking against the grain of traffic or rapid assembly/dispersion of groups of people), population and density measurements, and interactions between groups of people. This survey explores crowd analysis as it relates to two primary research areas: crowd statistics and behavior understanding. First, we survey methods for counting individuals and approximating the density of the crowd. Second, we showcase research efforts on behavior understanding as related to crowds. These works focus on identifying groups, interactions within small groups, and abnormal activity detection such as riots and bottlenecks in large crowds.Works presented in this section also focus on tracking groups of individuals, either as a single entity or a subset of individuals within the frame of reference. Finally, a summary of datasets available for crowd activity video research is provided.
Conference Paper
The medical research facilitates to acquire diverse type of data from the same individual for a particular cancer. Recent studies show that utilizing such diverse data results in more accurate predictions. The major challenge faced is how to utilize such diverse data sets in an effective way. In this paper, we introduce a multiple kernel based pipeline for integrative analysis of high-throughput molecular and clinical data. We apply the pipeline on Ovarian cancer data from TCGA. After multiple kernel have been generated from weighted sum of individual kernels, it is used to stratify patients and predict clinical outcomes. We examine the clinical outcomes of each subtype to verify how well they cluster.
Article
In the present paper, the avoidance behavior of pedestrians was characterized by controlled experiments. Several conflict situations were studied considering different flow rates and group sizes in crossing and head-on configurations. Pedestrians were recorded from above, and individual two-dimensional trajectories of their displacement were recovered after image processing. Lateral swaying amplitude and step lengths were measured for free pedestrians, obtaining similar values to the ones reported in the literature. Minimum avoidance distances were computed in two-pedestrian experiments. In the case of one pedestrian dodging an arrested one, the avoidance distance did not depend on the relative orientation of the still pedestrian with respect to the direction of motion of the first. When both pedestrians were moving, the avoidance distance in a perpendicular encounter was longer than the one obtained during a head-on approach. It was found that the mean curvature of the trajectories was linearly anticorrelated with the mean speed. Furthermore, two common avoidance maneuvers, stopping and steering, were defined from the analysis of the acceleration and curvature in single trajectories. Interestingly, it was more probable to observe steering events than stopping ones, also the probability of simultaneous steering and stopping occurrences was negligible. The results obtained in this paper can be used to validate and calibrate pedestrian dynamics models.
Chapter
The gathering of crowd traffic data either from videos or from visual observation has different uses. In the social simulation context, one of them is validating crowd behavior models and match the resulting traffic in control points with the real ones. When such models have been already validated, the immediate use can be aiding managers of facilities to infer, from real time data, what crowd behavior they should expect in their facilities. However, the transformation of those measurements into actual behavior patterns has not been satisfactorily addressed in the literature. In particular, most papers take into account a single measurement point. This paper contributes with an algorithm that produces possible populations that reproduces real traffic data obtained from multiple measurement locations. The algorithm has been validated against data obtained in a real field experiment.
Article
This paper proposes a novel approach for simulating pedestrian movement behavior based on artificial intelligence technology. Within this approach, a large volume of microscopic pedestrian movement behavior types were collected and encapsulated into an artificial neural network via network training. The trained network was then fed back into a simulation environment to predict the pedestrian movement. Two simulation experiments were conducted to evaluate the performance of the approach. First, a pedestrian-collision-avoidance test was conducted, and the results showed that virtual pedestrians with learned pedestrian behavior can move reasonably to avoid potential collisions with other pedestrians. In addition, a critical parameter, i.e., defined as “reacting distance” and determined to be 2.5 m, represented the boundary of the collision buffer zone. Second, a pedestrian counterflow in a road-crossing situation was simulated, and the results were compared with the real-life scenario. The comparison revealed that the pedestrian distributions, erratic trajectories, and density–speed fundamental diagram in the simulation are reasonably consistent with the real-life scenario. Furthermore, a quantitative indicator, i.e., the relative distance error, was calculated to evaluate the simulation error of pedestrians' trajectories between the simulation and the real-life scenario, the mean of which was calculated to be 0.322. This revealed that the simulation results were acceptable from an engineering perspective, and they also showed that the approach could reproduce the lane-formation phenomenon. We considered the proposed approach to be capable of simulating human-like microscopic pedestrian flow.
Article
Crowd understanding has drawn increasing attention from the computer vision community, and its progress is driven by the availability of public crowd datasets. In this paper, we contribute a large-scale benchmark dataset collected from the Shanghai 2010 World Expo. It includes 2630 annotated video sequences captured by 245 surveillance cameras, far larger than any public dataset. It covers a large number of different scenes and is suitable for evaluating the performance of crowd segmentation and estimation of crowd density, collectiveness, and cohesiveness, all of which are universal properties of crowd systems. In total, 53 637 crowd segments are manually annotated with the three crowd properties. This dataset is released to the public to advance research on crowd understanding. The large-scale annotated dataset enables using data-driven approaches for crowd understanding. In this paper, a data-driven approach is proposed as a baseline of crowd segmentation and estimation of crowd properties for the proposed dataset. Novel global and local crowd features are designed to retrieve similar training scenes and to match spatio-temporal crowd patches so that the labels of the training scenes can be accurately transferred to the query image. Extensive experiments demonstrate that the proposed method outperforms state-of-the-art approaches for crowd understanding.
Article
In this article we present an improved version of the Cellular Automata floor field model making use of a sub-mesh system to increase the maximum density allowed during simulation and reproduce phenomena observed in dense crowds. In order to calibrate the model’s parameters and to validate it we used data obtained from an empirical observation of bidirectional pedestrian flow. A good agreement was found between numerical simulation and experimental data and, in particular, the double outflow peak observed during the formation of deadlocks could be reproduced in numerical simulations, thus allowing the analysis of deadlock formation and dissolution. Finally, we used the developed high density model to compute the flow-ratio dependent fundamental diagram of bidirectional flow, demonstrating the instability of balanced flow and predicting the bidirectional flow behavior at very high densities. The model we presented here can be used to prevent dense crowd accidents in the future and to investigate the dynamics of the accidents which already occurred in the past. Additionally, fields such as granular and active matter physics may benefit from the developed framework to study different collective phenomena.
Conference Paper
The problem of modeling the navigation behavior of multiple interacting agents arises in different areas including robotics, computer graphics, and behavioral science. In this paper, we present an approach to learn the composite navigation behavior of interacting agents from demonstrations. The decision process that ultimately leads to the observed continuous trajectories of the agents often also comprises discrete decisions, which partition the space of composite trajectories into homotopy classes. Therefore, our method uses a mixture probability distribution that consists of a discrete distribution over the homotopy classes and continuous distributions over the composite trajectories for each homotopy class. Our approach learns the model parameters of this distribution that match, in expectation, the observed behavior in terms of user-defined features. To compute the feature expectations over the high-dimensional continuous distributions, we use Hamiltonian Markov chain Monte Carlo sampling. We exploit that the distributions are highly structured due to physical constraints and guide the sampling process to regions of high probability. We apply our approach to learning the behavior of pedestrians and demonstrate that it outperforms state-of-the-art methods.
Article
We present an online parameter learning algorithm for data-driven crowd simulation and crowd content generation. Our formulation is based on incrementally learning pedestrian motion models and behaviors from crowd videos. We combine the learned crowd-simulation model with an online tracker to compute accurate, smooth pedestrian trajectories. We refine the motion model using an optimization technique to estimate the agents׳ simulation parameters. We also use an adaptive-particle filtering scheme for improved computational efficiency. We highlight the benefits of our approach for improved data-driven crowd simulation, including crowd replication, augmented crowds and merging the behavior of pedestrians from multiple videos. We highlight our algorithm׳s performance in various test scenarios containing tens of human-like agents and evaluate it using standard metrics.
Article
In this paper, we propose the use of Tikhonov regularization in conjunction with node count as a general complexity measure in multiobjective genetic programming. We demonstrate that employing this general complexity yields mean squared test error measures over a range of regression problems, which are typically superior to those from conventional node count (but never statistically worse). We also analyze the reason that our new method outperforms the conventional complexity measure and conclude that it forms a decision mechanism that balances both syntactic and semantic information.
Article
Two evacuation model validation data sets collected as part of the EU FP7 project SAFEGUARD are presented. The data were collected from a RO-PAX ferry operated by Color Line (RP1) and a cruise ship operated by Royal Caribbean International (CS). The trials were semi-unannounced assembly trials at sea and involved some 1349 and 2500 passengers, respectively. The trials took place at an unspecified time; however, passengers were aware that on their voyage, an assembly exercise would take place. The validation data sets consist of passenger response times, starting locations, end locations, and arrival times in the assembly stations. The validation data were collected using a novel data acquisition system consisting of ship-mounted beacons, each emitting unique infrared (IR) signals and IR data logging tags worn by each passenger. The results from blind simulations using maritimeEXODUS for these assembly trials are presented and compared with the measured data. Three objective measures are proposed to assess the goodness of fit between the predicted model data and the measured data.
Article
Collective behaviors characterize the intrinsic dynamics of the crowds. Automatically understanding collective crowd behaviors has important applications to video surveillance, traffic management and crowd control, while it is closely related to scientific fields such as statistical physics and biology. In this paper, a new mixture model of dynamic pedestrian-Agents (MDA) is proposed to learn the collective behavior patterns of pedestrians in crowded scenes from video sequences. From agent-based modeling, each pedestrian in the crowd is driven by a dynamic pedestrian-agent, which is a linear dynamic system with initial and termination states reflecting the pedestrian’s belief of the starting point and the destination. The whole crowd is then modeled as a mixture of dynamic pedestrian-agents. Once the model parameters are learned from the trajectories extracted from videos, MDA can simulate the crowd behaviors. It can also infer the past behaviors and predict the future behaviors of pedestrians given their partially observed trajectories, and classify them different pedestrian behaviors. The effectiveness of MDA and its applications are demonstrated by qualitative and quantitative experiments on various video surveillance sequences.
Article
Crowd modeling and simulation (M&S) has been used to support the analysis of the behavior of crowds, in order to predict the impact of pedestrian movement and to test design alternatives. In recent years, crowd M&S has become more complex, and new technologies such as CAD (computer-aided design) and BIM (building information modeling) authoring tools are being used to support the process. There are challenges in adopting these technologies due to the lack of automation and integration of these tools for crowd M&S. We propose a method based on a distributed architecture with simulation in the cloud, and composition using workflows. In particular, we adopt a model-driven engineering approach to extract data from CAD/BIM authoring tools, Cell-DEVS theory for crowd modeling, simulation as a service to execute simulation remotely, and three-dimensional visualization. Finally, we present a case study for crowd evacuation, discussing the advantages of the proposed architecture. We show the advantages obtained when using distributed deployment, simulation-based design and collaborative development and we discuss how this facilitates the crowd behavior study and improves reusability in crowd M&S.
Article
Crowd modeling and simulation is an important and active research field, with a wide range of applications such as computer games, military training and evacuation modeling. One important issue in crowd modeling is model calibration through parameter tuning, so as to produce desired crowd behaviors. Common methods such as trial-and-error are time consuming and tedious. This paper proposes an evolutionary framework to automate the crowd model calibration process. In the proposed framework, a density-based matching scheme is introduced. By using the dynamic density of the crowd over time, and a weight landscape to emphasize important spatial regions, the proposed matching scheme provides a generally applicable way to evaluate the simulated crowd behaviors. Besides, a hybrid search mechanism based on differential evolution is proposed to efficiently tune parameters of crowd models. Simulation results demonstrate that the proposed framework is effective and efficient to calibrate the crowd models in order to produce desired macroscopic crowd behaviors.