Meet our CODE_n SPACES Resident blickshift! The spin-off from the University of Stuttgart develops software for efficient analyses of the eye movement behavior. Their product, in short, enables the automotive, marketing, and gaming industry as well as the usability research to develop new applications and services based on eye tracking technology. We talked to Michael Raschke, one of the three team members of blickshift about his vision, entrepreneurship in general and the decision to move into our CODE_n SPACES:
Let’s start out with the obvious: Can you explain, in lay terms, maybe with a couple of nice examples, what exactly you do?
Michael: Sure. Imagine you’re interested in finding out how people actually read an illustration in a newspaper. So you show ten people an illustration and ask them a question about it; then you draw their eye movements. A handful of standard techniques already exist for this type of analysis, and, in principle, you can use them as they are. But the moment you need to analyze the information in more detail and find out how specific individuals looked at the illustration, or whether there are groups of people who tend to look at it the same way, these techniques show their limitations. Or it takes an eternity to evaluate just a small experiment. We’ve developed a solution that sidesteps these disadvantages, allowing for extremely efficient eye movement analysis involving a large number of people. In essence, it’s based on the use of big data analytics in combination with highly interactive visualization techniques.
And who’s your solution for?
Michael: One example would be the car market. Imagine a situation like this: You’re driving down a road and you get to a really busy intersection with lots of people in cars and on motorcycles, and even pedestrians. And then there are road signs and traffic lights. Car companies want to know about drivers’ eye movements in such situations, so that they can evaluate them and adapt their driver assistance systems to better match the driver’s perceptions. The idea is to warn drivers about dangers they may not have noticed or perceived properly. So, for example, there might by a pedestrian who’s just about to cross the street and was almost overlooked. Car companies, their suppliers, and research institutions are carrying out tons of eye-tracking experiments on this at the moment.
This involves recording the eye movements of between 50 and 100 people in different situations, then analyzing the results and translating them into driver models that can be applied to vehicles further down the line. The analysis takes up a huge amount of time.
If you think about semi-automated or even fully automated driving, which the carmakers tell us will happen, they won’t just have to evaluate 50 or 100 guinea pigs, they’ll have to evaluate thousands, if not tens of thousands – just to allow for the development of extremely reliable driver assistance systems. Where we come in is when it comes to efficiently evaluating data from as many test drivers as possible, with just a few clicks of the mouse. At the moment, lots of things have to be done manually. Data has to be exported out of one application and imported into another. Lots of programs have to be written just to answer specific questions. This takes up a huge amount of time, so it’s very expensive. Also, it’s easy for errors to creep into the analysis. Our startup offers a solution to such problems. More…