The Big Interview: Dr Angelika Kneidl, accu:rate
Following the deaths of 21 people at the Love Parade in Duisburg, Germany, during a crush in the crowd, Dr Angelika Kneidl felt that she had a solution at her fingertips.
Working at the time on a PhD on innovative crowd movement technologies, she began receiving telephone calls asking whether her team’s software could help prevent other similar disasters, when hundreds of people were crushed in a tunnel leading in and out of the venue.
“We were receiving calls asking whether our work could help with planning these events, especially regarding crowd safety. By the time I had finished my PhD, my company was set up ,” Angelika said. “We were bringing something new in terms of our algorithms to crowd simulation modelling that hadn’t been seen in Germany at that time.”
Only three years on and with its innovative crowd:it software, Munich-based accu:rate has become one of the go-to businesses in Germany and is spreading its wings across Europe.
The models they use, she said, come not from the more traditional floating particle physics, but from studying human behaviour and human steps, the “optimal steps model”.
It looks at different ways humans think, with different goals and motives and can simulate those wishing to get to their target as quickly as possible, as well as those who want to go slower.
She said: “Our model represented a step-change in functionality compared to what came before it, and we saw a significant improvement in results in comparison to old standard models. It has been proven successful with less than a 5% difference in evacuation time compared to video footage when we tried it on a university evacuation.
“It makes it more precise and allows us to look at some of the randomness of human behaviour, such as someone suddenly stopping to buy a bar of chocolate.
“We are working with a museum at the moment and we have to simulate so many different people and factors: stopping at an exhibit or not, going to the toilet, just going quickly round, lots of things.
“Everyone is different, with different thought processes, and we need to recreate that in order to build a successful model.”
Angelika said that businesses were still focussed on crowd simulation as a safety tool rather than an experience one, but that the technology had so much to offer, and has huge opportunities before it.
“Safety is still so huge in Germany,” she said. “And that has always been the main consideration. But I think people are starting to understand the magnitude of what the software can do, what it can tell us and how it can help in the planning stages; not just for safety but for overall customer experience.
“We have the technologies, but it requires businesses to buy into their capabilities and not just use them for what they have always used them for.
“Businesses across Europe, I think, are still a little behind in their thinking on this, but are slowly starting to see the value besides safety.”
One increasingly popular area, she said, was in sensor technology and how it can be used to improve buildings during operation. However, she said that given the “hype” surrounding them, “I’m not sure they can deliver what they are promising or whether they are really ready yet.”
A new impluse of the sector is using Artificial Intelligence and machine learning to speed up the testing and adjustment of algorithms against real-world data, so that the machine can learn from real-data and adapts accordingly.
“Uncertainty quantification allows to identify the most influencing and important parameters form the variety of parameters needed for crowd simulations. A study of the 2005 London bombing (Prof John Drury) showed survivors not just running away, but helping other victims. The influence on the overall evacuation time was significant. But whether 60, 70 or 80 percent of the victims have been helpful was not relevant to the overall evacuation time. So, even if we do not know for sure the exact numbers, we are still able to reproduce realistic results, if we take into account the behavioural patterns.”
With Germany lagging behind the UK and other countries, and having made a strong name for themselves, accu:rate is looking abroad to more mature markets that are already embracing the technology.
“It’s huge around major events these days, and wherever in the world major events are taking place, that is where new crowd simulation technology can help improve the planning and operation.
“And there’s definitely a strong future for modelling in the building planning process. Rather than addressing whole cities in one go, we are more concentrated on individual buildings as we think that if you concentrate on the detail of the buildings one by one, it will help the city grow by itself. It’s about starting at the bottom and working up.”