Is AI the Best Solution for Crowd Management?

We increasingly use technology for a broad variety of purposes, and the more that happens, the more data we collect and store. These days, AI is transforming the way we utilize that information. Machines can read and learn from different types of data and then perform real-world tasks. 

That’s true in almost every sector, but AI is also being used to simplify and improve how humans control crowds and populations worldwide. From cities to events and the pandemic, tech has transformed the way we’re managing interactions and gatherings.

While AI can augment, automate, and improve decision-making and planning, it’s only a part of the equation. Machines can interpret images from cameras and traffic or crowd movements and behaviors, but there’s still a lot of human work to do when it comes to crowd management.

Photo by Franki Chamaki on Unsplash

AI and Crowd Control

So, where does AI get employed in crowd management, and how does it work? Typically, machines learn to interpret things like crowd counts and movements by reviewing thousands and thousands of carefully selected images. 

Engineers will use a diverse selection of pictures and angles to teach machines about things like perspective and distinguishing humans from other objects. After that, images get converted into pure data, and algorithms are used to make determinations around any given requirement. A series of yes/no decisions lead the AI to come to a specific conclusion and act in a certain way. 

Human security specialists can program Ais to automate parts of their roles – like counting crowds in stadiums at sporting events and predicting problems or monitoring social distancing measures in many types of environments like shopping malls or at the beach, for instance. 

During the pandemic, AI can even be employed to track crowd movements in high foot traffic areas, recording which surfaces get touched, whether people are wearing masks, and a range of other factors.

Finland using AI and cameras to monitor crowds

Finland has been conducting practical research into using AI for everything from regional population planning to crowd control. Local authorities and the police force are active in assessing the viability of using AI when hosting things like music festivals and sporting events, and Helsinki tested a system during August and September of 2020, employing AI across existing video surveillance equipment in the city. The primary aim of the project is to develop methods of assisting authorities with deployment and monitoring. Essentially, the technology would be used to alert police well in advance of crowd numbers in any given sector exceeding safety requirements. 

Finland’s Digital and Population Data Services Agency is also trialing AI to manage population migration and resources allocation, monitoring demographic changes in urban and rural areas. 

AI and Crowd Management in India

Crowd management issues at Indian melas are nothing new. These religious gatherings attract thousands of people, and stampedes are common. In 2015, the local government in Andhra Pradesh approached the Lab for Spatial Informatics for assistance with managing a huge amount of imaging from CCTV cameras – their purpose was to monitor foot traffic in different sections or ‘ghats’ while the festival was underway. The team adapted tech from a recent road traffic project to provide headcounts and crowd density management. 

The solution outperformed traditional police headcounts in a hectic setting, and the team estimated a 75% accuracy rate. Data was relayed to a website where attendees could find out about crowd density in different sectors of the gathering in real time. Police also utilized the live information to redirect pilgrims to less crowded ghats and decide where to allocate resources and personnel.

Physical Crowd Control 

All the many examples of AI-enhanced crowd control out there share one common aspect – they complement humans, not replace them. While AI can perform an important role and automate some of the tasks involved with policing and monitoring crowds, security basics are still necessary. 

Murray Bilby, Crowd Control Expert at Todoos, says that with a solid security plan in place, AI-enhanced measures can help your own event go smoothly. “It’s important to start security for any event well in advance of commencement. Begin by notifying local emergency services and authorities of your intentions and select a venue that can handle the visitor numbers you expect. Don’t even think about opening the doors until you have the means to communicate with attendees – like a PA system. Expect the unexpected because crowds can get out of control fast, and when they do, you need to be able to talk over the din.”

Bilby adds that signage should be clear, practical, and highly visible wherever there’s a crowd. People should know where to line up for entry and exit, facilities should be marked out to avoid unnecessary movement, and attendees should instantly know about any off-limits areas and the locations where they can find medical assistance. “Once done with signage, it’s essential you lay out the venue properly. Use barricades and stanchions, cones, ropes, and tapes to mark areas and walkways. Generally, people will do what they’re supposed to do when they know the rules.”

“Another aspect of crowd control where tech can reduce problems is with ticketing and checking in. These days, you can minimize entry times by using QR codes and online ticketing. With the right mix of planning, person-power, and utilization of tech, there’s no reason your event can’t happen safely and smoothly.”