Week 8 Lecture - Part 2 PDF Human Factors in Built Environments

CVEN4405 Human Factors in Civil and Transport Engineering Week 8 Lecture – Part 2 Human factors in built environments, crowds and emergencies Dr Milad Haghani School of Civil and Environmental Engineering Week 8: Human factors in built environments, crowds and emergencies An empirically informe...

CVEN4405 Human Factors in Civil and Transport Engineering Week 8 Lecture – Part 2 Human factors in built environments, crowds and emergencies Dr Milad Haghani School of Civil and Environmental Engineering Week 8: Human factors in built environments, crowds and emergencies An empirically informed modelling framework Decision-making layers Econometric models Locomotion layers Models from statistical mechanics, physics and transportation Local pathfinding Local pathfinding The social force model of crowd motion An empirically informed modelling framework Back to where we were: Simulation & modelling Visualisation of Modelling Bottleneck parameters are by far the most important parameters, followed by decision adaptation parameters. Let’s revisit two fundamental papers Let’s revisit two fundamental papers The origin of the notion of “faster-is-slower” idea What we call the faster-is-slower effect is more like a fallacy than a fact https://www.sciencedirect.com/science/article/pii/S096585641831111X?dgcid=author#m0045 Slow moving crowd Relatively competitive crowd Very competitive (fast moving) crowd Calibrating the social force parameters Friction parameter and relaxation-time parameter Calibrating the social force parameters Friction parameter and relaxation-time parameter What we call the faster-is-slower effect is more like a fallacy than a fact The origin of the notion of “obstacle” effect Helbing, D., Farkas, I.J., Molnar, P. and Vicsek, T., 2002. Simulation of pedestrian crowds in normal and evacuation situations. Pedestrian and evacuation dynamics, 21(2), pp.21-58. The idea of barriers facilitating crowd flow at bottlenecks is also more like a fallacy than a fact Other limitations of the Social Force Model • The social force model cannot handle navigation, wayfinding, and decision-making, and was not originally designed to so either. • Since the notion of “force” is this model are metaphorical and not real, the measurement of these force is obviously impossible, making a full calibration of this model very difficult. We need to resort to heuristic methods (just like the model itself which can also be considered a heuristic model). To this date, I am not aware of a established and universally accepted methodology for calibrating the social force model parameters. • But I can say for sure, that, some parameters of this model are more important than other parameters. So, instead of a full calibration, we may resort to a partial calibration focusing on those critical parameters. Summary notes on simulation modelling • The accuracy of a crowd model is far more dependant on the rigour of its parameter calibration than its theoretical construct. • Any plausible theoretical construct can potentially produce accurate enough results, should it undergo a proper calibration procedure. • Major modelling flaws are not always visually detectable by inspecting the plausibility of the visualisation of simulation process. • Poorly calibrated (or even worse, uncalibrated) models can produce misleading and counterproductive recommendations. • When it comes to modelling accuracy, some parameters are far more important than others. Priority should be given to them. • A first test of a model accuracy should always be “how accurately it reproduces bottleneck flow rates” • We need more observational data and more detailed methods of calibrating bottleneck-related parameters in crowd dynamics research. • Theoretical advancements in modelling cannot be considered acceptable without certain degrees of connection to empirical data (of some sort, field or experimental). That’s basically what I stand for in this research field. • Realistic looking avatars and visualisation does not necessarily mean a good (trustworthy) model! A bad model is sometimes worse than having no model at all. The way Crowds are described in academic literature Panic Irrationality Herding Why did these terms become so ubiquitous in the crowd dynamics literature so much? Panic (N=37) Herding (N=10) What’s the origin of this trend? Answer: The trend started emerging mainly following Helbing et al (2000) publication in Nature How prevalent have these terminologies been? Answer: Very prevalent 2019 Is there a gold solution? • • Crowds are sources of trouble and unruly behaviour and their irrational behaviour is often then cause of incidents. Is it computer simulation? Is it reinforcement and control? The old-fashioned paradigm Semantics matter. When an emergency strikes, information should be withheld from crowds, otherwise they will panic. Faster-is-slower assumption The faster-is-slower assumption https://arxiv.org/abs/2310.20506 https://www.straitstimes.com/asia/east-asia/crowd-density-seen-as-likely-cause-of-south-korea-tragedy Stampede or crush? Why semantics matter here https://twitter.com/Reuters/status/1586493967637819392?s=20&t=1Vm0aYKo_mGha2uuD089wg https://tnp.straitstimes.com/news/world/halloween-crowd-crush-seoul-was-absolutely-avoidable-experts-say https://youtu.be/TAjFeBDIjJs?si=1RrWY6vQefvX5Ufj https://youtu.be/_7LTm_1LNto?si=us8wB8OB0TE2ka1Y https://youtu.be/282cmirDwNk?si=_Sv4VSDrYgRrvXJI Other lab-in-the-field experiments How people make exit decisions How people make decisions within social groups (reaction decision, exit choice etc) How people make a decision to change their decision Other lab-in-the-field experiments How people decisionmaking is affected by observing peer behaviour How people decisionmaking is affected by elevated level of stress/urgency Other lab-in-the-field experiments How crowd behaves at bottlenecks How people resolve conflicts while on the move What have we learnt from these experiments? • People are capable of making complicated evacuation decisions and trade-offs even in fractions of seconds • They do not make their decisions based on a single factor. They are capable evaluating a range of attributes • People do not necessarily follow each other, like a herd. Their decision-making is more complex than that. What have we learnt from these experiments? What have we learnt from these experiments? • People dynamically revisit and change their decisions during evacuations • When we increase the level of urgency/stress, their decisions do not become irrational, rather more optimal What have we learnt from these experiments? What have we learnt from these experiments? What we call the “faster-is-slower” effect is more like a fallacy (we will talk about this further) What have we learnt from these experiments? The mechanism of decision-making is different when people are in groups compared to individuals What have we learnt from these experiments? Group decision can be made through consensus What have we learnt from these experiments? • Group decision can be made through leadership What have we learnt from these experiments? Group decision can be made through conformity to majority What have we learnt from these experiments? What have we learnt from these experiments? • A lot of what we have learnt from old-fashioned computer simulations do not hold under experimental scrutiny • A lot of what is believed generally about crowd behaviour (from a few pioneering influential papers) is not observable/replicable under experimental conditions. • It is possible to “model” and “predict” people’s behaviour using experimental data. • It is even possible to learn crowd behaviour from surveys and questionnaires! • Results seem replicable and show signs of (convergent) validity. • Data/model from one experimental condition is capable of replicating independent observations (external validity). The past and existing landscape of empirical crowd dynamics literature Empirical methods in crowd dynamics are getting increasingly popular https://www.sciencedirect.com/science/article/pii/S0378437121004180 The evolution of crowd dynamics literature 1980-2020 Interventional approaches: A knowledge gap in crowd research Mathematical optimisation Architectural optimisation Behavioural optimisation The Swiss Cheese Model Reason, J. (1997). Managing Risks of Organizational Accidents. Ashgate Publishing Limited, Aldershot, UK Download the paper via this link: Source: Haghani, Milad and Coughlan, Matt and Crabb, Ben and Dierickx, Anton and Feliciani, Claudio and van Gelder, Roderick and Geoerg, Paul and Hocaoglu, Nazli and Laws, Steve and Lovreglio, Ruggiero and Miles, Zoe and NICOLAS, Alexandre and O’Toole, William and Schaap, Syan and Semmens, Travis and Shahhoseini, Zahra and SPAAIJ, RAMÓN and Tatrai, Andrew and Webster, John and Wilson, Alan, Contemporary challenges in crowd safety research and practice, and a roadmap for the future: The Swiss Cheese Model of Crowd Safety and the need for a Vision Zero target (May 8, 2023). Interventional approaches: A knowledge gap in crowd research Mathematical optimisation Architectural optimisation Behavioural optimisation From a descriptive to a prescriptive/interventional approach Is herding behaviour beneficial or detrimental? An example of behavioural optimisation Herding in exit choice making Herding in exit choice changing Is decision adaptability behaviour beneficial or detrimental? An example of behavioural optimisation Is waiting strategy more efficient than instant response? An example of behavioural optimisation High degree of waiting Moderate degree of waiting Instant response No waiting No pre-movement delay Is waiting strategy more efficient than instant response? An example of behavioural optimisation Instant reaction Moderately heterogenous reaction Highly heterogenous reaction Can the crowd benefit from a mixture of strategies? An example of behavioural optimisation Follow the majority strategy Follow the minority without possibility of decision change strategy Follow the minority with possibility of decision change strategy Do we need to modify the behaviour of the entire crowd in order to reach efficiency? From a descriptive/observational to a prescriptive/interventional approach The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments my while analyses are ongoing and publications pending. The new generation of crowd evacuation experiments: Behavioural optimization experiments (UNSW, 2022) Kindly please be mindful of the Intellectual Property of the design of these experiments while my analyses are ongoing and publications pending. Any safety educational program or campaign should be evidence-based (as opposed to fiction based) https://www.youtube.com/watch?v=W-BBfcbQX7Q&t=94s I conclude that, there is no panacea solution to the crowd safety problem, but a holistic multi-layered safety system that utilises active participation of all potential stakeholders can significantly reduce the likelihood of disastrous accidents. At a global level, we need to target a Vision Zero of Crowd Safety, i.e., set a global initiative of bringing deaths and severe injuries in crowded spaces to zero by a set year.