How trams are becoming smart and autonomous – Interstars

Driver assistance systems are now mandatory in many new cars. But these are not only necessary for cars, but also for trams. At Vision, Automation and Control Center au Austrian Institute of Technology (AIT) they have been looking for him for 8 years.
This year, a research project was completed in which assistance systems for “Bims” with artificial intelligence (AI) have been equipped. “Tomorrow’s trams will also have to develop anticipatory behavior in order to operate safely and avoid breakdowns”, specifies the AIT project manager. Christian Zinner in conversation with futurezone.

driving assistance

This is done with neural networks that learn to understand traffic situations and assess them with foresight. This is mainly to avoid bodily injury. The boxes integrated into the trams are equipped with an AI supposed to understand and interpret delicate situations. “People are and will remain the measure of all things,” says Zinner.

“But people don’t have the same level of attention all the time because drivers sometimes get tired when they’re on the road for a long time. This is where we can offer assistance with AI,” explains the AIT project manager. “There are always accidents due to driving errors, for example because a vehicle is driven too quickly around a bend and then derails or overturns,” says Zinner. Here too, the AI ​​should help to avoid them.

How AI recognizes vehicles and speeds

Installation in fleets

The new components used in the FFG research funding project »INTELLiTRAM“ are used must also be as small as possible because they must be able to be integrated into existing fleets. It was a challenge, said Michel Kreilmeiergeneral manager of Integrated Missionwhich is an AIT project partner with Alstom.

Trams also have special requirements that do not exist for cars: the systems must know signals and turnouts, for example. In addition, the braking distances are much longer, adds Kreilmeier. “The AI ​​helps to better interpret the scenes. »

The tram could enter the depot independently if AIT is successful

Autonomous driving to the depot

The project also aims to allow the tram to run partially autonomously in the future. But can it happen soon? “We don’t expect all vehicles to be driverless overnight. But there are areas of application where it is relatively easy to imagine it being used in trams as well,” says Zinner.

One of these fields of application concerns the deposit. The tram learns to park there independently and reactivates itself when it is time for it to leave the depot. Overnight, vehicles are usually parked at the depot and not returned to service until the start of the shift. It is very expensive for companies if they have to pay night supplements. The pumice stone should also be able to handle the transfer to the workshop or car wash independently, if AIT is successful. To do this, you will need to bring the necessary technology directly into the vehicle.

Besides the conversion of fleets, there are other challenges. The more autonomous the systems are, the more security requirements there are. “It’s even bigger Theme standardization and certification“, explains Kreilmeier. That is why he does not dare to make fixed predictions as to when autonomous trams will actually be used in fleet management.


A “Smart trams thanks to Sense, Learn and React” (abbreviated: INTELLiTRAM) are these Austrian Institute of Technology (AIT) with Europe’s largest rail vehicle manufacturer Alstom (formerly Bombardier) and Integrated Mission involved

La science

Researchers can build on the results and use them further

financial aid

The project lasted 3 years and was financed within the framework of the Mobility of the Future program by the Federal Ministry for Climate Protection sponsored and by Austrian Research Promotion Agency colonized

assistance systems

AIT has already developed two assistance systems for trams that could serve as the basis for the project. “Our driver assistance system monitors how fast the car is going to a certain point and whether the intended speed is maintained,” says Zinner. If this is not the case, audible warnings are given to drivers. If the limit is greatly exceeded, the vehicle is automatically braked.

A second assistance system relates to collision avoidance. “To do this, we have developed an obstacle system with stereo cameras which is integrated into the car”, explains the project manager. Both of these streetcar assist systems are ready for the market and are in use in several cities around the world, Zinner says. Frankfurt in Germany was the first city to use these systems Zurich followed in Switzerland. “Penetration in tram fleets will increase rapidly,” says Zinner.

search strategy

“For us, the project is part of a larger research strategy,” says Kreilmeier of Mission Embedded. Work continues there to ensure that AI systems pass a so-called “safety check” in order to be approved. The AI ​​systems developed are currently not approved for use in trams. AIT then studies how to better recognize the rails in order to be able to maintain them in a timely manner.

This series is published independently with the financial support of the Research Promotion Agency (FFG).