BLOG: 5 questions for Robin Janssens
R&D is at the core of CrowdScan. We are very happy to have Robin Janssens on board. About one year ago he joined us for writing his Ph.D. handling different ways to optimize our system and make better use of the data.
"Neural networks have huge potential, especially for monitoring crowds in challenging environments like subway or railway stations."
1. How did you end up at CrowdScan?
After finishing my master’s degree in applied engineering at the University of Antwerp, I started as a research engineer at IDLab, the academic research group where CrowdScan technology was invented. When looking for new research funding, CrowdScan started taking off and they were looking to expand their team. We decided to apply for industrial Ph.D. funding. After writing a project proposal and defending it in front of an expert panel, we got awarded the project.
2. What is your Ph.D. all about and when is it supposed to be finished?
The project includes different parts. One thing we are trying to achieve is to optimize the system so it can make better use of the data. We also want it to combine with other data and sensors to be able to get more crowd metrics and improve the accuracy of the overall system. The project started at the beginning of 2022. Ph.D. projects typically have a timespan of 4 years.
3. You are now working for 1 year for CrowdScan. How do you look back at this year? What are your main accomplishments till now?
One of the main projects I have been working on is to improve the reliability of our counting solution for public transportation environments. Using a technology that is based on radio waves can be challenging in environments such as subway or railway stations, where large metal structures, like trams, are being introduced which reflect and absorb the radio waves in often unpredictable ways. Typically filtering out such noise would be done on processed data. However, for public transport environments, we typically want to capture also fast changes in crowd movements like on platforms, making filtering very difficult without affecting the data. Currently, we are harnessing the capabilities of neural networks to understand the impact and make adjustments to the raw, unprocessed data to compensate for them. This gives us the possibility to filter out anomalies without affecting the quality of the actual information stream. We are already deploying this technology for multiple public transport providers across Europe.
4. You are working with neural networks? What is that exactly and why is this so promising for CrowdScan?
When using neural networks we can make our algorithms learn more complex relationships between data, thereby enabling us to derive insights that would be unattainable through the use of simpler algorithms. The goal is to allow for more reliable systems that are easier to work with. Neural networks also have huge potential in other applications such as forecasting and improving early warning systems.
5. In your opinion, what are the main advantages of CrowdScan (technology)?
CrowdScan's radio wave technology is one of the only technologies on the market that can measure crowd density directly. In particular in open and dynamic environments, this technology stands unparalleled in terms of its accuracy and versatility. Crowd density information is the foundation of any crowd analytics application. This can be supplemented with additional information from sensors or third-party information sources based on the customer's needs.