Much of the news about the imminent arrival of autonomous cars is focused on the car manufacturers, but much of the technology that will allow an automobile to navigate autonomously will be produced by tech companies outside of the auto industry.
For example, satellite images from Google Maps seem to be good, but they carry very little detailed information. By contrast, images from airplanes are far more detailed: road angles and houses can be recognised on a road up to an accuracy of two centimeters. However, if you want to locate all the traffic lights, construction and petrol stations, you need a lot of time, patience and manpower – or a machine can take over the work. This is exactly what the Munich startup Terraloupe is trying. With artificial intelligence, the company wants to speed up the process significantly.
The company uses an algorithm to record the course of the road and the location of road signs. The objects are identified by a deep-learning technology and recorded on the map, explains founder Christian Schaub. Google relies on a similar technology to geo-capture cities and landscapes. The Start-up is still far from the same level as the Google, but its own technology is already sufficient to generate its own business model, according to its founder. These much more accurate and detailed maps are interesting for many industries.
Detailed map information can, for example, be used for the navigation of autonomous vehicles. A self-driving car must not only perceive immediate obstacles, but must also know about traffic signs and road conditions.
“Data on transport infrastructure is of interest to automotive manufacturers, suppliers and car services,” says Schaub.
Some major publicly listed companies in Germany are already among the customers of the Munich company, including customers from the energy and insurance sectors. An insurer is, for example, interested in whether a solar installation is located on a house in order to be able to make a building evaluation. On the basis of this information, the insurance contribution can be more accurately measured, explains Schaub.
Terraloupe was founded in 2015 by Manuela Rasthofer, Christian Schaub and Sebastian Gerke and now employs 25 people. Rasthofer comes from the military sector, where she previously made simulations from aerial photographs, says Schaub, who was previously with Siemens. The other co-founder, Gerke, researched deep-learning algorithms. Terraloupe is now Germany’s first company, which has linked both areas and applied such algorithms to aerial images, says Schaub.
Recently, Terraloupe has received a seven-figure financing from the affiliate Bayern Kapital, as Wiwo founders reported. With the capital, Terraloupe wants to hire additional developers and improve the algorithm, says Schaub.
“Entering the US market is also planned,” says the founder.