and Transport Company VIP Postdam: On the path to an autonomous

siemens.tld/keyword 9pt

Fully automated and driverless On the path to an autonomous tram A partially or fully autonomous tram, airport rail systems from Siemens normally operate in a complex on the other hand, has the advantage have long been operating in environment where other road users that it doesn’t tire, it can react faster Chicago, for example, as well as like cars, bicycles, and pedestrians are than a human, and it has a more in metros in Paris, Taipei, and constantly on the move. Traffic lights extensive field of perception. Sensors Nuremberg. Val systems have oper- and signaling systems regulate traffic replace eyes and ears, and pattern ated in Lille since 1983 followed at critical points. Driving a tram basi- recognition, basic behavioral patterns, by other cities, and a new fully cally requires driving by sight at pre- and reactions need to be incorporated automated cityval system will be scribed speeds with scheduled stops into the tram control programs. How- inaugurated in Rennes in 2020. along the line, monitoring the board- ever, human intuition and experience ing and departing of passengers, and Automated operation is primarily can be standardized only to a limited quickly dealing with any unusual suitable for closed-rail systems like extent, and not every situation is situation or emergency. A tram driver metros and mass-transit systems predictable. The tram control system has to be able to handle at least 25 running on completely isolated rail needs to be gradually programmed so different tasks, many of them simulta- lines. Autonomous operation in an it can react flexibly to each new situa- neously. These include everything open infrastructure, as is typical of tion. The software has to learn to from reacting to car drivers, pedestri- urban trams, requires completely make decisions at lightning speed – ans, and trams ahead, maintaining set different technologies: An autono- or request support. speeds and schedules under all weath- mous tram has to learn to function er conditions, stopping precisely at intelligently in a complex environ- traffic lights, signals, and passenger ment. stops, coordinating connections, and Because mature infrastructure in responding quickly and appropriately urban areas can be altered very little, to any problems that occur in the tram the technology needs to be installed in or along the line. The driver ultimately Human operators still aboard: the tram itself. And the technology is responsible for the well-being of the Testing semi-autonomous tram must be able to handle even the most passengers. With experience, a driver operation in the depot and on the complex traffic situations occurring can intuitively master all of these line in with a Siemens with mixed traffic. situations. Combino.

Autonomous Combino tram Information from the cameras is On all test routes, the tram from in Potsdam continuously combined with data from Siemens Mobility GmbH stops precise- Transport company ViP Verkehrs­ the three radar detectors. The reflect- ly at each station platform and re- betrieb Potsdam GmbH has been a ed radar waves identify items like sumes its journey on its own when customer of Siemens Mobility since metal objects such as cars and other allowed by the signal. People ap- the 1990s and has provided a Combino trams and enable the speed and proaching the track quickly are warned tram that’s been equipped for autono- distance of these objects to be pre- by the tram bell before the tram ap- mous driving tests with GPS, comput- cisely calculated. Three LiDAR scanners plies its brakes, as usual. The tram ers, and several types of sensors, operate as digital eyes to scan and reacts to cars crossing the tracks just including cameras, laser (LiDAR), and capture a 270° image of the space like a driver would. There are no radar scanners. The joint development around the front end of the Combino. jerking movements, no sharp braking project will study the technological They scan horizontally and vertically ahead of curves, and no leisurely challenges of autonomous driving and can detect people, measure their acceleration, all of which indicate that under real-life conditions in order to speed, and calculate and position their the tram is operating autonomously. develop and test viable solutions. The movement in a stylized three-dimen- Testing on the extended route and on world’s first test vehicle was presented sional image of the detected surround- automated depot runs are proving that at InnoTrans 2018, but it wasn’t de- ings. Complex algorithms running on the autonomous control system can be signed for commercial operation. Over powerful computers are the “brain” of used successfully with the skills it has the long term, however, Siemens the tram. They analyze the real-time learned. Mobility GmbH will develop a tram traffic situation, predict its ongoing that can operate at the GoA 3 level of development, and make the right Future challenges automation (accompanying a driver to decisions that are then performed by The ability to recognize obstacles and handle emergencies) or the GoA 4 actuators for the Combino’s bell, drive signals should soon be so advanced level (without accompanying opera- system, control electronics, and brak- that the system can be used in normal tors). ing system. tram operations. Because there are currently no approval procedures for The space ahead of and to the sides of The route first needs to be learned by autonomous rail vehicles, the Associa- the tram is captured by an array of the tram. Test runs have been con- tion of German Transport Services and cameras. The system’s analytic soft- ducted in Potsdam since May 2018 Siemens Mobility are working together ware is trained with methods of artifi- accompanied by a driver who inter- even at this early stage to define the cial intelligence to recognize countless venes only in the event of danger. necessary safety and legal framework objects and people of any shape and in Information processed in real time for the new technology. every position. Among other things, from data provided by the cameras the algorithms recognize signals and and sensors is displayed on a large The sensors used on the test tram signal conditions, people, other ve- monitor in the test tram. The test came from the industrial and automo- hicles, and objects and obstacles on route has been extended from the tive sectors. Different conditions for the tracks ahead. Camera images are original six to 13 kilometers and now pricing and durability apply in those analyzed and stop the tram when runs from the ViP depot to the final areas and for the specific operating traffic lights or signals block further stop at Marie-Juchacz-Strasse. Buses conditions of a tram. Although Sie- movement. When the light or signal also travel along an 800-meter-long mens Mobility is drawing on existing changes to “go,” the Combino auto- roofed and fenced stretch where automotive technologies, it’s adapting matically resumes its journey. various scenarios can be tested. the basic pattern recognition and signal processing algorithms to the requirements of tram operations. Automotive sensors, for example, record many more objects in front of and alongside a car than are relevant for a tram. To correct this, stationary objects like power masts, fences, and street markings that are no danger to the tram must be filtered out in the data processing. Deriving appropriate warning and action strategies poses another chal- lenge. Although the components have algorithms useful for recognizing people, only a specialist can develop algorithms that recognize when a person presents a danger. In pedes- trian zones and at tram stops, for example, people move very close to the danger zone of a tram but must be assessed completely differently than in the case of a car. The research prototype of a smart autonomous tramway has been successfully tested under real road traffic conditions on a six-kilometer section of the Potsdam tram network

In addition to these new functions for Another potential option is to run line the Siemens assistance systems, the sections with a separate track on an first intermediate levels of automated automated basis, which would elimi- operation will be achieved in the next nate the need to perfectly master few years. The first commercially participation in mixed traffic with useful step is being planned: automat- other road users. That would take the ed depot operation using an autono- burden off the driver and provide mous tram. This means that time- improved safety, while programmed intensive marshaling processes in the driving behavior, which would also be Matthias Hofmann depot – like service runs through a able to take account of trams ahead, Matthias is Product Portfolio Manager washing system to the storage track – would save energy on these sections. for Tramways and Systems at Siemens Mobility GmbH, and since can be performed automatically. Because the sensors in the tram pre- 2015 he has been responsible for the Because the depot is largely separate cisely register the tram’s surroundings, development program for assisted from the rest of the public transporta- they can also be used to assist the and autonomous tramway operation. tion system, this simplifies technical driver in a conventional tram or to He has worked in various positions controls and eases approval from monitor the condition of the infra- involving light rail at Siemens since authorities. The project is called AStriD structure for maintenance purposes. 1994, including International Sales (from the German for “autonomous Automated procedures can continu- and Marketing, Portfolio Manage- tramway in depot”) and is sponsored ously check the clearance gauge and ment, and Project Management. by ’s Federal Ministry of condition of the tracks and overhead Transport and Digital Infrastructure lines and warn of fallen trees or other (BMVI) in accordance with the Moder- obstacles on the track. nity Fund (“mFUND”) funding guideline. They can also monitor the integrity of The project is being implemented at signals, points, and balises and see the Potsdam Transport Services depot. where vegetation is intruding onto the From the very start, the development tracks, for example. In all cases, the process will take into account the legal new technology helps the driver and provisions governing approval and provides greater safety, higher avail- operation of an autonomous tramway ability and improved passenger com- system and the economic framework fort, punctuality, and energy savings. in which this system could operate. Fewer accidents, less wear and tear, The AStriD project was launched in and lower repair costs also reduce Translation of article published in October 2019, with a planned duration operating costs and ensure a lasting of three years. „bahn manager Magazin“, edition 02/2020, increase in value throughout the entire https://www.bahn-manager.de/archiv/ lifecycle. © Siemens 2020