Crowd Egress Research Articles

A2X: An end-to-end framework for assessing agent and environment interactions in multimodal human trajectory prediction

In recent years, human trajectory prediction (HTP) has garnered attention in computer vision literature. Although this task has much in common with the longstanding task of crowd simulation, there is little from crowd simulation that has been borrowed, especially in terms of evaluation protocols. The key difference between the two tasks is that HTP is concerned with forecasting multiple steps at a time and capturing the multimodality of real human trajectories. A majority of HTP models are trained on the same few datasets, which feature small, transient interactions between real people and little to no interaction between people and the environment. Unsurprisingly, when tested on crowd egress scenarios, these models produce erroneous trajectories that accelerate too quickly and collide too frequently, but the metrics used in HTP literature cannot convey these particular issues. To address these challenges, we propose (1) the A2X dataset, which has simulated crowd egress and complex navigation scenarios that compensate for the lack of agent-to-environment interaction in existing real datasets, (2) evaluation metrics that convey model performance with more reliability and nuance, and (3) a guideline for future data acquisition in HTP. A subset of the proposed metrics are novel multiverse metrics, which are better suited for multimodal models than existing metrics. The dataset is available at: https://mubbasir.github.io/HTP-benchmark.

A systematic review and meta-analysis on the effect social groups have on the egress times of pedestrian crowds

Conflicting results in the literature raise the question of how reliable findings from single experiments on pedestrian crowd movements are. An important example is the effect social groups have on crowd egress times from confined spaces where both increases and decreases were reported. We identify only six comparable studies and conduct the first analysis on this topic that integrates evidence from multiple studies quantitatively, accounting for different sample sizes. The aggregated findings suggest social groups increase average egress times but there is insufficient evidence to reject the null hypothesis of no effect. The conflicting results across published studies are thus likely to have arisen by chance, as experiments are statistically underpowered for determining a small effect. We find no evidence for publication bias in terms of findings or statistical power. Our work presents a quantitative basis for discussing the statistical reliability of experiments considering the high context-dependency of pedestrian dynamics.

Verifying the applicability of a pedestrian simulation model to reproduce the effect of exit design on egress flow under normal and emergency conditions

Crowd egress at narrow exit is a popular research topic, due to its intrinsic importance in architectural designs and building codes. However, relatively few studies have been conducted to verify the performance of pedestrian models for crowd escape at exits, especially relating to different exit designs. This paper aims to verify the applicability of a microscopic pedestrian simulation model, Social Force Model (SFM), embedded in Viswalk software to reproduce the effect of exit design on egress flow under normal and emergency conditions. Empirical data from controlled experiments considering the effects of obstacles size and location of exits under normal and emergency conditions were tested and compared with the simulation from the SFM. Results indicated that after parameter optimization, Viswalk simulation model can provide reasonable estimates for crowd escape under normal situations with a mean RMSE value 1.97s for total evacuation time. However, the simulation model was less capable in reproducing the emergency condition. As compared to the empirical data, clogging events were less spotted under emergency in the simulation. Faster-is-slower effects were not found in both empirical and simulation scenarios. In addition, the exit location effects from simulation data agreed with empirical data, corner exits were more efficient than middle exits under both situations. Meanwhile, the obstacle effects, as observed in empirical data, were less reproduced in the simulation, especially under emergency conditions. The results suggest that the application of the Viswalk model in simulating emergency situations needs scrutiny and further investigations in the future with empirical data.

Predicting Crowd Egress and Environment Relationships to Support Building Design Optimization

Evaluating and optimizing the design of built and yet-to-be-built environments, with respect to human occupancy and behavior is both greatly beneficial and challenging. Crowd simulation can provide the computational means to analyze a design through the movement of virtual occupants (agents). A range of analytic information (metrics) can be computed from the simulated movement of the agents that offer insights on the design. Crowd simulation and the related analysis can be part of interactive or offline design optimization pipelines. Unfortunately, large scale crowd simulations are prohibitively expensive, especially when used within iterative design and optimization loops, where hundreds of simulations often need to be computed at interactive rates. We propose a machine learning framework that aims to solve this problem by learning the relationship between a building design and the evaluation metrics extracted from expensive simulations. We train an offline regression neural network using a synthetic training set that we generate for this purpose. Once the network is trained it can evaluate new designs efficiently, and approximate the corresponding analytic information with high accuracy. The proposed framework can also be used to find an optimized layout. We demonstrate the effectiveness of the framework on a variety of real world case studies.

Evaluation of Manual Traffic Control in Directing Traffic Operations for Major Special Events: Clemson University Football Game Day Experience

Crowd ingress and egress occurring before and after large special events results in oversaturation of the traffic network and is a crucial issue event planners and transportation officials must address. Because intersections constitute the greatest restraint in network traffic flows, efficient operation during special event peak flows is crucial. This research focuses on evaluating how well traffic control officers optimize intersection operations during heavily oversaturated conditions associated with large-scale special events. Network traffic data, including volumes, queues, and traffic officer signal/right-of-way phase times, were collected before and after four Clemson University home football games during the 2014 and 2015 season. Actual traffic count volumes were adjusted to account for vehicle queues and used to develop optimal signal timings using Synchro. These results are compared to field-observed intersection operation using officer directed hand-signaling and officer traffic signal pushbutton operation. A microscopic VISSIM model was also created for both manual control and optimized control scenarios to determine average delays for each approach and to improve the interpretation of macroscopic Synchro results. Findings indicate that traffic officers perform well in near saturated conditions; however, optimized signal timings provide reduced approach delay and overall intersection delay in heavily oversaturated conditions. The paper includes recommendations on how traffic officers can improve intersection performance. The paper also describes how results from data collection, analysis, and simulation modeling have been used to make recommendations to law enforcement in evaluating traffic operations and identifying improvements. Intersection control is crucial to the implementation of the overall traffic management plan, and the research findings provide insight to the effectiveness of law enforcement control of intersection operations.

Insights toward efficient angle design of pedestrian crowd egress point bottlenecks

Understanding the impact of geometrical characteristics of bottlenecks in pedestrian egress behaviour is ubiquitous in safe and effective crowd management. This paper studies how funnel-shaped bottlenecks with different angles affect the microscopic and macroscopic properties of pedestrian egress flow. The minimum width of the bottleneck in all experiments was one metre and the egress point was exactly in the middle. Two scenarios were designed and experimented. First, the large width of the bottleneck was constant and the length, and subsequently the slope, was varied. Second, the length of the bottleneck was constant and the large width, and subsequently the slope, was varied. As a result, the impact of length, large width, and slope were examined. The results suggest that the best-performing bottleneck with regard to angle follows a certain trend, with 26.6° being the most efficient. Further, it was observed that upon increasing the large width the flow increases.

Pedestrian crowd dynamics in merging sections: Revisiting the “faster-is-slower” phenomenon

The study of the discharge of active or self-driven matter in narrow passages has become of the growing interest in a variety of fields. The question has particularly important practical applications for the safety of pedestrian human flows notably in emergency scenarios. It has been suggested predominantly through simulation in some theoretical studies as well as through few experimentations that under certain circumstances, an elevated vigour to escape may exacerbate the outflow and cause further delay although the experimental evidence is rather mixed. The dimensions of this complex phenomenon known as the “faster-is slower” effect are of crucial importance to be understood owing to its potential practical implications for the emergency management. The contextual requirements of observing this phenomenon are yet to be identified. It is not clear whether a “do not speed up” policy is universally beneficial and advisable in an evacuation scenario. Here for the first time we experimentally examine this phenomenon in relation to the pedestrian flows at merging sections as a common geometric feature of crowd egress. Various merging angles and three different speed regimes were examined in high-density laboratory experiments. The measurements of flow interruptions and egress efficiency all indicated that the pedestrians were discharged faster when moving at elevated speed levels. We also observed clear dependencies between the discharge rate and the physical layout of the merging with certain designs clearly outperforming others. But regardless of the design, we observed faster throughput and greater avalanche sizes when we instructed pedestrians to run. Our results give the suggestion that observation of the faster-is-slower effect may necessitate certain critical conditions including passages being overly narrow relative to the size of participles (pedestrians) to create long-lasting blockages. The faster-is-slower assumption may not be universal and there may be circumstances where faster is, in fact, faster for evacuees. In the light of these findings, we suggest that it is important to identify and formulate those conditions so they can be disentangled from one another in the models. Misguided overgeneralisations may have unintended adverse ramifications for the safe evacuation management, and this highlights the need for further exploration of this phenomenon.

Evaluating and Optimizing Evacuation Plans for Crowd Egress.

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.

Understanding differences in emergency escape and experimental pedestrian crowd egress through quantitative comparison

Models of pedestrian crowd behavior in panic or emergency situations rarely have complementary empirical data to calibrate or validate the model's prediction. Till date, there is no systematic study that has attempted to conduct a quantitative comparison of emergency escape with the laboratory experimental data for the underlying differences in crowd behavior.In this paper, an attempt is made to explore the differences between experimental data on pedestrian crowd egress and emergency egress at a bottleneck and their implication for crowd safety. The laboratory experiments consist of pedestrians passing through a bottleneck of varying widths. The emergency egress data has been extracted from the publicly available rare video footage of panic escape at a shopping center during an Earthquake in China. The extracted data are then used to test a force-based crowd simulation model.It was found that there are underlying differences on the crowd egress behavior (headway distribution and flow) under emergency conditions as compared to controlled laboratory experiment. There is significant number of group exits and less lone exits in emergency egress. It was observed that the force-based model was able to replicate the normal egress as well as panic egress. It is recommended that any pedestrian crowd evacuation simulation model be tested to reproduce this fundamental headway distribution as observed in this study which includes >85% of headway as short headway (≤0.5s) with predominately group exits (0–0.1s headway) rather than lone exits for the bottleneck widths between 1.7 and 2.1m.

Human exit choice in crowded built environments: Investigating underlying behavioural differences between normal egress and emergency evacuations

Egress behaviour of pedestrians in crowded complex confined spaces is investigated in this study. Despite recent methodological progress in the development of simulation tools for predicting crowd egress and evacuation, little is known based on empirical data about the underlying rules that govern exit wayfinding of pedestrians in multi-exit places. Particularly, fundamental differences between behavioural features of emergency and non-emergency egress have not been fully explored by previous studies. Stated-choice data was collected in face-to-face interviews with passengers as they exited a major railway station in Melbourne. Participants were asked what exit decision they would have made given certain hypothetical scenarios at that same station. Econometric models (error-component mixed logit) were developed to quantify the way passengers evaluate and prioritise various contributing factors while accommodating the potential decision heterogeneity. These factors include distance, crowding, visibility of exits, proximity of the exits to their destination, impact of other passengers’ decisions; and spatial distribution of exits. Key findings of our modelling suggest that (1) for nonemergency egress, proximity of the exit points to passenger's destination is a dominant factor although not the sole determinant. (2) In an emergency, passengers place a much higher priority on avoiding crowded exits compared to non-emergency situations. (3) Directional flows of pedestrians do not significantly impact on decisions made in a normal egress. (4) In an emergency evacuation, directional flows are considered as a negative utility factor by majority of individuals, although the perception of this factor is highly heterogeneous and also depends on the visibility of the exits targeted by the flow. The proposed static models can be incorporated with a broad range of crowd simulation methods as alternatives for heuristic modeller-defined exit choice rules. They accommodate the dissimilarities of egress behaviour between emergencies and non-emergencies, thereby enabling planners to conduct virtual simulations under either scenario.

Conceptualizing intragroup and intergroup dynamics within a controlled crowd evacuation.

Social dynamics play a critical role in successful pedestrian evacuations. Crowd modeling research has made progress in capturing the way individual and group dynamics affect evacuations; however, few studies have simultaneously examined how individuals and groups interact with one another during egress. To address this gap, the researchers present a conceptual agent-based model (ABM) designed to study the ways in which autonomous, heterogeneous, decision-making individuals negotiate intragroup and intergroup behavior while exiting a large venue. A key feature of this proposed model is the examination of the dynamics among and between various groupings, where heterogeneity at the individual level dynamically affects group behavior and subsequently group/group interactions. ABM provides a means of representing the important social factors that affect decision making among diverse social groups. Expanding on the 2013 work of Vizzari et al., the researchers focus specifically on social factors and decision making at the individual/group and group/group levels to more realistically portray dynamic crowd systems during a pedestrian evacuation. By developing a model with individual, intragroup, and intergroup interactions, the ABM provides a more representative approximation of real-world crowd egress. The simulation will enable more informed planning by disaster managers, emergency planners, and other decision makers. This pedestrian behavioral concept is one piece of a larger simulation model. Future research will build toward an integrated model capturing decision-making interactions between pedestrians and vehicles that affect evacuation outcomes.

Animal dynamics based approach for modeling pedestrian crowd egress under panic conditions

Collective movement is important during emergencies such as natural disasters or terrorist attacks, when rapid egress is essential for escape. The development of quantitative theories and models to explain and predict the collective dynamics of pedestrians has been hindered by the lack of complementary data under emergency conditions. Collective patterns are not restricted to humans, but have been observed in other non-human biological systems. In this study, a mathematical model for crowd panic is derived from collective animal dynamics. The development and validation of the model is supported by data from experiments with panicking Argentine ants ( Linepithema humile). A first attempt is also made to scale the model parameters for collective pedestrian traffic from those for ant traffic, by employing a scaling concept approach commonly used in biology.

An Anticipative Crowd Management System Preventing Clogging in Exits During Pedestrian Evacuation Processes

This paper presents an anticipative system which operates during pedestrian evacuation processes and prevents escape points from congestion. The processing framework of the system includes four discrete stages: a) the detection and tracking of pedestrians, b) the estimation of possible route for the very near future, indicating possible congestion in exits, c) the proposal of free and nearby escape alternatives, and d) the activation of guiding signals, sound and optical. Detection and tracking of pedestrians is based on an enhanced implementation of a system proposed by Viola, Jones, and Snow that incorporates both appearance and motion information in near real-time. At any moment, detected pedestrians can instantly be defined as the initial condition of the second stage of the system, i.e., the route estimation model. Route estimation is enabled by a dynamic model inspired by electrostatic-induced potential fields. The model combines electrostatic-induced potential fields to incorporate flexibility in the movement of pedestrians. It is based on Cellular Automata (CA), thus taking advantage of their inherent ability to represent effectively phenomena of arbitrary complexity. Presumable congestion during crowd egress, leads to the prompt activation of sound and optical signals that guide pedestrians towards alternative escaping points. Anticipative crowd management has not been thoroughly employed and this system aims at constituting an effective proposal.

Animal dynamics based approach for modeling pedestrian crowd egress under panic conditions

Collective movement is important during emergencies such as natural disasters or terrorist attacks, when rapid egress is essential for escape. The development of quantitative theories and models to explain and predict the collective dynamics of pedestrians has been hindered by the lack of complementary data under emergency conditions. Collective patterns are not restricted to humans, but have been observed in other non-human biological systems. In this study, a mathematical model for crowd panic is derived from collective animal dynamics. The development and validation of the model is supported by data from experiments with panicking Argentine ants (Linepithema humile). A first attempt is also made to scale the model parameters for collective pedestrian traffic from those for ant traffic, by employing a scaling concept approach commonly used in biology.

Crowd circulation and stadium design: low flow rate systems

For the past 30 years, sports stadia design guides have contained generalised optimistic flow rates for the design of crowd evacuation and egress routes. Unfortunately, even though the guides recognise their own limitations and the fact that lower flow rates should be considered in certain situations, human nature, practical needs and a lack of research has lead to widespread use of guidance values that are sometimes inappropriate. It is imperative within risk assessment and circulation design that this phenomenon is better understood in order to limit the potential for underdesign at the risk of unwanted levels of queuing, congestion and potentially unsafe conditions. This is especially important when considering the progressive move from prescriptive guidance to performance based design within the design industry. In recognition of this, the new European Standard for the design of spectator facilities recommends design flow rates significantly lower than those recommended in traditional sports guidance documents. This paper presents a significant contribution to this work by firstly providing supportive evidence for the new European standard values when applied to football and rugby stadia, and secondly stating that the figures within traditional and new guidance should not be generally applied to what amounts to over 15 different stadia types. Case study evidence is provided to show that the new European standard values still remain up to twice those observed in some stadia types. Reasons are given for the occurrence of low-capacity flow rates within these stadia and recommendations are made for future work to ensure that designers are using appropriate design flow rates for each stadium project.