Showing foresight in surroundings detection: strategic partnership between Knorr-Bremse and Israeli start-up RailVision.

Interview with Elen Katz, Founder and CEO of the Israeli start-up RailVision , about groundbreaking solutions for surroundings detection, autonomous driving as a trending topic in the rail industry and the strategic partnership with Knorr-Bremse

Mr. Katz, RailVision solutions are able to detect obstacles on railway lines up to two kilometers away - even at night and during fog, rain and storms.

RailVision offers by far the most modern surroundings detection technology for rail vehicles on the market. Our unique selling point is the ability to detect obstacles at greater distances. Our concept is based on an intelligent combination of video and infrared technology. The use of infrared cameras, in particular, is quite special in this field of application. One of the tools we are working on is digital maps, so that ultimately our system will be familiar with every aspect of the route, every signal, every piece of infrastructure. The real-time surroundings detection system will then make any deviations from these conditions immediately visible. In the same way, our products can detect damage to tracks, switches or overhead lines even before they fail. So our support also extends to the timely and continuous maintenance of rail infrastructure. Our unique hardware solution and cutting-edge AI software positions us as a leader in our domain. We use vision sensors that were optimized for our performance, which provides unique results. Our technology has its roots in the military sector, but all RailVision products are intended exclusively for civilian use.

RailVision offers by far the most advanced surroundings detection technology for rail vehicles on the market.

Elen Katz – Founder and CEO of the Israeli start-up RailVision

How did you come up with the idea of using this technology in rail transport?

We knew that the rail sector worldwide suffered from a considerable number of delays and accidents caused by obstacles on lines. Upon closer inspection, it became clear to us that human capabilities and technical necessities are often quite at odds. A young, healthy train driver can see around 200 meters in daylight and good visibility, and perhaps 80 meters at night with spotlights. At higher speeds, however, a train often needs a braking distance of over half a kilometer to come to a stop. By the time the train driver has spotted an obstacle by eyesight, there is often no chance of stopping the train in time. This small calculation illustrates that humans are, in many cases, the Achilles heel in the rail sector. Rail operators only make money from their trains when they are running. Our original goal was therefore to use our technology to reduce costly downtimes in rail transport.

Is that no longer the right approach?

We also recognized that the rail sector faces another significant problem in many countries: there are fewer and fewer railway workers. Newspapers report, for example, that the Austrian Federal Railways will have to fill 8,000 jobs in the next five years, 2,000 of which will be locomotive drivers. In Switzerland, 750 train drivers will retire in the next few years; there are virtually no applicants to replace them. In many cases, the young generation is no longer interested in working in this industry. Besides train drivers, this goes for workers who maintain the infrastructure, including the thousands of kilometers of rail network.

RailVision’s solutions offer detection and classification of objects or obstacles (e.g. humans, vehicles, signals), rail path recognition (i.e. switch state detection), distance measurement and opportunity infrastructure condition monitoring, required for ATO.

Surroundings detection and autonomous driving as key topics for the railway industry

Can technology provide solutions to the problems mentioned above?

Today, the railway industry – like many other branches of transport and logistics – is compelled to consider how to make use of electronics, computer technology and robotics. Autonomous driving (known as ATO, automatic train operation, in railway jargon) has been on everyone’s lips in the rail industry and a prominent theme at all railway fairs. However, autonomous driving only becomes possible with professional and reliable surroundings detection.

Are you alone in the market?

No, but we are certainly the market leader with our technological approach, i.e. combining video and infrared technology with artificial intelligence. We are several years ahead of our competitors, including some large corporations.In the vehicle sector, radar and laser technology are used to detect obstacles. RailVision operates with electro-optics.

In the vehicle sector, radar and laser technology are used to detect obstacles. RailVision operates with electro-optics.

We use high-resolution video cameras that see hazards alongside infrared devices that can "feel" obstacles based on temperature differences. This combination generates very vivid and informative images. Radar and laser technology are active systems, which means beams have to be emitted before being reflected back by any obstacles. These systems provide an enormous amount of information that needs to be filtered and analyzed by computers. That takes a lot of processing power. In addition, the amount of iron usually used in rail infrastructure like rails, overhead lines, signals etc. often causes interference in radar systems. Laser technology also comes up against its limits sooner in the rail sector. You cannot simply beam into the eyes of waiting passengers as the train enters a station, for instance. Cameras and infrared devices, on the other hand, are passive systems that provide reliable data material when intelligently combined. And their images can be analyzed far more quickly than radar images, for example.

Sensor technology to meet the demands of trains

Most of your competitors are originally from the automotive industry.

Yes, that's why their sensor technology is based on the demands of road traffic. This involves braking distances of 50 to 80 meters – much shorter than on rail tracks, where 200 to 800 meters is a normal range. Unlike all our competitors, we have exclusively addressed the specific demands of trains, which is why we are the market leader in the industry today. There are currently some manufacturers on the market offering surroundings detection for short-distance trains such as trams and light rail systems. We have no competition in the mainline rail sector. As things stand, we can use our cameras to perform route and environment detection at far greater distances than our competitors.

So long as the railroad section is straight…

Railroad curves are adapted to the speeds of the trains. The faster the train travels, the larger the radius of the curve has to be, otherwise the train would come off the rails. Our ability to perceive long distances maps very well onto this principle.

In a few years your system will apparently be able to detect obstacles up to four kilometers away.

Yes, we would like to rapidly develop this "far-sightedness". Especially in the high-speed segment, where we have braking distances of up to four kilometers. That's why we're putting all our available brainpower into this work. Orders we have already received for field tests in the area of autonomous train operation show that we are on the right track. As part of a two-year project with the Swiss Federal Railways, we are testing our technology in ten locomotives used in freight and marshalling yards.

Knorr-Bremse as a financially strong partner with an outstanding reputation

How did the cooperation between RailVision and Knorr-Bremse come about?

Like all start-ups, we were looking for financially strong partners, ideally close to our industry. We made two attempts to talk to Knorr-Bremse, but at first they were not interested. About a year ago, we suddenly managed to get in touch with the management and were fortunately given the opportunity to demonstrate our innovations. This was the start of a "love story" that led Knorr-Bremse to acquire 21.3% of RailVision's shares in March 2019 with the aim of establishing a strategic partnership. In a funny way, you could call us the sugar babes, offering know-how and creative fun with technology, and Knorr-Bremse the sugar daddy, bringing its capital, years of industry experience, access to customers and experience in safety technology systems.

Who holds the other 78.7 percent?

The founders of the company, some private investors and a public company from Israel that mainly focuses on surroundings detection in road traffic. This is our largest investor. Knorr-Bremse, as a strategic investor, is our sole partner in the rail transport sector.

What is the benefit of this strategic partnership for RailVision?

We are absolute pioneers when it comes to our technology. On the other hand, we do not yet know the railway industry very well and are not well established in terms of production and sales. Knorr-Bremse possesses this comprehensive knowledge as an experienced company with representatives and customer bases all over the world, a strong financial background and, aside from its expertise in safety-related technology, an outstanding reputation.

Israel – creative pool for cyber security and internet start-ups

Israel has not yet stood out as a hotspot for the international rail industry.

I had the idea in 2004 to develop this system in my home country. To this day Israel has no rail industry; there is not a single company that produces even screws or seats for railways. That is why it was incredibly difficult to raise funding for this project in the beginning.

Israel has an excellent reputation as a creative pool.

Yes, but under tremendous competition from cyber security and internet start-ups, it was immensely difficult for us to convince financiers to even look at our project. Finally, this public company that works with obstacle detection in its sector listened to us. When their managers took a closer look at the international rail market and found that there was no similar technology anywhere, they provided an initial two million dollars of capital.


Dr. Nicolas Lange, Member of the Management Board of Knorr-Bremse Rail Vehicle Systems, explains how and why Knorr-Bremse plays an active role in Digital Business and cooperation with start-ups:

  • The idea of engaging systematically with start-up collaborations and the topics of Digitalization and Digital Business was born at Knorr-Bremse in 2017. To this end, a cross-divisional "Digital Board" was set up at the highest management level to regularly exchange information on developments, goals and projects in these areas.
  • Since then, they have been on the lookout for fresh and compatible business ideas - with the help of scouts, collaborations with startup accelerators such as "TechFounders", hackathons, their own "Pitch Days" and trade fairs. At the same time, start-ups are approaching Knorr-Bremse independently.
  • Knorr-Bremse's goal is to cooperate with start-ups or idea pools that develop innovative ideas with great agility and energy in small teams – ideas that may in turn complement our business. Such collaborations may lead to investment, as in the case of RailVision, or they may not. We are also paying increasing attention to the USA and Asia with a view to gaining additional know-how that meaningfully complements our existing competences, products and business ideas. In return, start-ups gain Knorr-Bremse as a partner: a competent global player to help them scale their business.
  • RailVision was the very first start-up that Knorr-Bremse partnered with. The Israeli company has developed surroundings detection systems for rail vehicles based on video and infrared technology. This is a promising field in terms of realizing autonomous train operation. In March 2019, Knorr-Bremse signed an agreement to invest USD 10 million in RailVision, acquiring a 21.3 percent stake in return. In addition, Knorr-Bremse offers practical support in areas such as marketing, customer contacts and industry know-how.
Back to overview