Artificial intelligence will help prevent train accidents and increase train punctuality.


Both passenger and freight trains are already preparing for the introduction of fully autonomous trains. At the moment, driverless trains work only on separate lines, and autonomous driving systems are used in subways in e.g. Paris, Sydney and Shanghai. These solutions are developed thanks to artificial intelligence, which is now used worldwide in high-speed railroads. Although in Poland they are only at the planning stage, AI technologies are already being developed and implemented by Polish carriers and engineers. The benefits of intelligent solutions include, among others, increased safety and prevention of breakdowns, reduced costs, improved train punctuality and passenger comfort.



“In January this year, the Council of Ministers adopted the document specifying the policy for the development of artificial intelligence in Poland for the coming years. One of the priorities is the development of autonomous vehicles. AI is especially needed in high-speed railroads, which are only at the planning stage in Poland, therefore at the moment there is no direct stimulus for the development of this technology. However, many companies run projects on applying artificial intelligence in railroad transport, both passenger and freight,” says Aleksander Lisowiec, the Head of the Department of Intelligent Networks at the Łukasiewicz Tele and Radio Research Institute.


The government's "Policy for the development of artificial intelligence in Poland from 2020" points out that this technology has great potential to boost the national economy. The development of AI in Poland is expected to annually increase GDP dynamics by approximately 2.65 p.p., and by 2030 automate up to 49% of the working time in Poland, while generating better paid jobs in key sectors. Transport is one of them, including rail transport, which is among the industries most susceptible to the benefits of AI implementation. However, there are some barriers to its development in Poland.


“Rail transportation is a very broad field, and while artificial intelligence brings financial benefits in the long term, a capital expenditure is necessary, first. There are also very strict safety requirements in rail transportation. There is some resistance to replacing the human decision maker in critical situations with a machine. Barriers to the development of artificial intelligence in railroads include legal standards: according to Polish law, there must be at least one driver in a rail vehicle, which limits the development of autonomous vehicles,” Aleksander Lisowiec explains.


Before driverless autonomous trains appear on the Polish tracks, intelligent technologies will be used, for example, to manage railroad stock. Sensors integrated with AI algorithms installed in trains will be able to collect, analyse and provide data on current location, distance made and speed of the train. This enables operators to increase or decrease frequencies of trains depending on the needs and number of passengers, reduce costs and improve travel comfort.


“Artificial intelligence helps manage trains in various ways. First and foremost, it is transport logic, which means adjusting the frequency to the volume of passenger traffic, making sure there are no stoppages, punctuality, safety of travelers and maximum reduction of train travel time,” says the expert of the Łukasiewicz Research Network.


One of the artificial intelligence systems already used in rail transport is the so-called eco-driving system. It is based on adjusting the speed of a moving train so that it uses as little electric energy as possible.


“Artificial intelligence algorithms are based on data provided by sensors installed in the railroad infrastructure: in stations, on tracks and in cars. This data includes e.g. train speed and position. This is how, among others, the European Rail Traffic Management System (ERTMS) works. Video camera-based systems are also widely used,” Aleksander Lisowiec explains.


One of the main goals of implementing AI-based solutions on the railroads is to increase safety. According to data from the Railway Transport Office, there were 425 accidents on the railroads in 2020 (including six serious accidents). This is a decrease of 19 percent compared to the previous year, however it is mainly caused by the pandemic — in 2020, passenger and freight trains traveled a total of nearly 15 million "train-kilometers" less than in 2019. On the other hand, as every year, the highest number of accidents at railway-road crossings: 170 accidents were reported last year (compared to 199 the year before).


The use of AI-based systems would avoid many such accidents. At train speeds of 200–300 km/h, the driver is generally not able to see all the dangers on the track in advance. Artificial intelligence may help. It also helps detect a breakdown or wear in a component early enough that could lead to an accident, such as a train derailment.


“Such a case may happen due to a broken wheel, which may be caused by a seized bearing. To prevent this, temperature sensors are installed close to the wheel bearing. Data from these sensors is transmitted to the central unit, which, based on analysis, may immediately stop the train, or plan an inspection of the car at the right time," explains an expert of the Łukasiewicz Research Network – Tele and Radio Research Institute. "In the case of a freight train, derailment may also be caused by the shifting of loads in the car, and here, again, sensors may be installed to examine the condition of suspension. When tilting of the car is detected, the train is also stopped.


One of the other common failures involves blocked brake shoes, which leads to wheel damage. Sensors integrated into the AI allow monitoring the condition of the car's brakes and preventing such incidents.


“At our institute, together with Meritus Systemy Teleinformatyczne, we run a project to create a system consisting of a network of sensors installed in a railroad carriage. These sensors measure the basic driving parameters of the carriage and transmit the data to a central unit. Based on this, the unit is able to predict the possibility of a failure. If such a failure threatens to damage the car or derail the train, the train is immediately brought to a standstill,” Aleksander Lisowiec says.


The sensors are in the form of small "boxes" attached with a magnet at various points in the car. The data from these sensors is then transmitted to a central unit that collects it and sends it to a cloud-based server using GSM. In addition, the central unit also determines the position of the car using signals from GPS and Galileo satellite navigation systems, allowing the carrier to monitor its location online. This is particularly important for freight operators, who can then, for example, provide their customers with information on planned loading and unloading times.


The second institute from the Łukasiewicz Research Network — KOMEL also develops solutions in the field of highly advanced data communications techniques for the safety of railroad traffic. This is to serve the created by engineers telemetric recorder of critical parameters TEREPAK used in rolling stock. The device allows for remote monitoring of important drives, measurements and analysis of operational parameters of selected devices. Thanks to the application of communication via the Internet or GSM, TEREPAK enables remote control of the system, observation and comparison of parameters from various sensors and also allows for notification of emergency situations to the services responsible for the operation of machines.

Warning: Use of undefined constant cookiesinfo - assumed 'cookiesinfo' (this will throw an Error in a future version of PHP) in /home/platne/itrorgpl/public_html/itr/en/index.php on line 105