Christian Studer CPO and Co-founder of Xovis.

Enter the AI sensor generation

Anina Jampen

Xovis CPO Christian Studer on how Xovis enters the AI sensor generation.

Released in 2017, the PC2R will be upgraded with AI-powered firmware in June 2018. At the same time, Xovis rolls out the all-new PC2S sensor. What can we expect?

The new Xovis PC2S Sensor with integrated AI marks the beginning of a new generation of 3D sensors. Of course, we wanted to give that capability to our latest sensor, the PC2R, too. In combination with the Xovis SPIDER processing unit, our customers get the true full-area monitoring and customer journey analytics they always wanted. Starting with this release, the PC2S and PC2R sensors will apply AI-powered algorithms that will especially spur the performance for analyzing large areas with lots of challenging situations.

This sounds like a lot of initial installation effort.

No, in fact, the installation process for the new PC2S is going to be even more “plug-and-play.” The need for masking error-producing areas of the field of view by hand is reduced because of the outstanding data intelligence on the sensor itself. This simplifies the sensor setup and reduces labour cost for our partners. False counts are minimized from the beginning, improving the data accuracy for every single sensor. Most importantly, setting up a Multisensor becomes more comfortable and results in more accurate tracking data optimizing the rollout of full-area monitoring solutions.

How exactly will the AI-powered Xovis algorithms boost in-store analytics?

In the retail industry, sensors and algorithms help improve the understanding of customer behaviour. Currently, 3D sensors deliver a rather trivial set of KPIs, even though very accurately. Systems count entering and exiting visitors or measure dwell times in certain areas of the store. Sensors with integrated AI collect more granular and more robust data, increasing the data accuracy of these KPIs. The data gathered with a high precision AI-powered people counting and tracking technology removes the hurdles for in-store analytics.

Could you elaborate on that?

With reliable 3D sensors, it is relatively easy to determine if an object of the shape and height of a person is moving in a particular direction within a store. In dense areas, accuracy nevertheless leaves room for improvement. That is where AI-powered 3D sensors excel in reducing false counts. Accordingly, we can take full-area monitoring and customer journey analytics to the next level. Retailers could for instance identify hot and cold spots based on customer tracks or measure waiting times to avoid queue quitters.

What other developments are we going to see in the near future?

The whole AI-powered in-store analytics open the door for future dimensions of customer behaviour analysis. In combination with the Xovis SPIDER, our unique processing unit, we want to empower brick-and-mortar stores to keep up with the online competition. Especially, in the age of stores with automated, unstaffed check-out facilities, AI-powered 3D sensors are vital. They detect, when a person grabs an article from a shelf, puts it back or makes a buying decision. We believe that the recognition of such behavioral detail is the key to taking cashier-free stores from the trial stage to becoming a global standard.

So basically, Xovis puts the focus on providing a full-fledged retail solution?

We deliver the tools for that, yes. Nevertheless, the industry has proven to be relatively complex; Sure, the required data for sophisticated in-store analytics needs to be based on AI-powered Xovis 3D Sensors. But for proper analytics and consulting, we will always cooperate with our competent partners in the retail sector. We see ourselves as a provider of data intelligence, not data analytics. Our primary focus, therefore, remains on algorithmic research and in-house sensor manufacturing.