Artificial intelligence in public transport

Taking public transport to the next level - achieving this will require an interplay of many digitization technologies and innovations: Automation, Artificial Intelligence (AI), Big Data, as well as standardization and interoperability within the industry will enable buses and trains to form the backbone of smart mobility. With a clear focus on passenger needs and user-friendliness, these innovations will make public transport an even more essential part of sustainable mobility.

The basis of artificial intelligence is data of which there is plenty. Whether historical or real-time traffic data, data from ticketing, timetable planning or fleet management, the collected data is so detailed and extensive that it can be used to train algorithms. Applying AI to this data can reveal changes and trends more precisely and more quickly. It enables improved departure predictions, optimal route planning and rapid responses to current events such as accidents or traffic jams.

Significantly improved departure predictions thanks to machine learning

In a pilot project between INIT and Golden GateBridge, Highway & Transportation District in San Francisco, for example, an AI-based software solution from INIT’s Austrian subsidiary inola, significantly improved the accuracy of bus departure predictions, in the most extreme case from 49 to a remarkable 87.47 percent. ML-Core, inola's software product, provides predictions based on historical data and real-time information. It uses innovative algorithms for pattern recognition and can process large amounts of data (Big Data). In MOBILEstatistics, our evaluation and statistics system, the operating data is collected and processed with lots of additional information like driving times. Based on this data, the inola ML-Core has various methods at its disposal for automatic pattern recognition and forecasting. The software recognizes and uses the best method for each scenario. If the prediction quality deteriorates, the ML model is automatically retrained. In this way, changes in traffic flow or line changes are immediately detected and mapped. This would not be possible with conventional forecasting methods.

Based on the trained model, the ML-Core calculates driving time predictions for all trip sections. The newly predicted departure times for stops are then compiled from these individual values and transmitted to subsequent systems. Current traffic restrictions or the driving time of the vehicles immediately ahead are also taken into account. Passengers can view live predictions, making it even easier to plan their journeys, which ultimately increases passenger satisfaction.

INIT's MOBILE-ITCS nextGen includes ML forecasting as standard.

AI for more accurate occupancy predictions and increased passenger comfort

Knowing how full a bus or train will be allows passengers to adjust their travel plans accordingly and switch to a less busy connection if necessary. This requires reliable real-time information on occupancy levels, as provided by MOBILEguide, our occupancy information and passenger guidance system. The technological challenge lies in providing a solid database to determine a vehicle’s expected occupancy rate. MOBILEguide uses state-of-the-art technology and sophisticated algorithms based on cutting-edge system architecture. If vehicles are not equipped with counting sensors, then other data sources are used, such as Wi-Fi and Bluetooth signals from smartphones or requests to connection information and the Mobilithek platform of the German Federal Ministry for Digital and Transport, an open data portal that provides mobility, geo and weather data.

The current occupancy rate is determined on the basis of the transferred boarding and alighting data after each departure from a stop, linked with timetable data and checked for plausibility. This provides an actual measured number of passengers in a vehicle. Using the vehicle capacities, the current load is calculated and stored.

A patented process and an AI algorithm are used to predict how busy a vehicle will be. This is done by matching real-time occupancy rates in the back-office system with typical boarding and alighting behavior at the next stop, which is obtained from historical data. In this way, the expected number of alighting passengers is also taken into account. INIT’s solution outperforms conventional systems in terms of reliability. 

The information can then be displayed for the dispatcher in the Intermodal Transport Control System MOBILE-ITCS, for example. In addition, occupancy information can be distributed via passenger information channels, for example via apps or the transport companies‘ websites. This allows passengers to adjust their travel plans accordingly and, if necessary, switch to less crowded connections, which in turn ensures a more evenly distributed passenger load.

AI research

The topic of Artificial intelligence is an integral part of numerous research projects to help us in the further development of our products and solutions. For example, the project KARL (Competence Center KARL - Artificial Intelligence for Work and Learning in the Karlsruhe Region) assesses the impact of artificial intelligence on the work environment and company organizations. The aim is to design and test human-centered, transparent and AI-assisted work and learning systems. As part of KARL, INIT is researching AI-based assistance systems to support control center personnel. The AI is meant to be trained in such a way that it takes into account numerous factors such as historical situations and then proposes dispatching measures that are precisely adapted to the current situation.

The DaKiMo research project aims to demonstrate how mobility data can be processed with the help of artificial intelligence and enriched with additional data, such as traffic or weather conditions, to create intelligent information services for citizens and the economy. The aim is to promote resource-efficient, sustainable mobility in these times of ever-growing climate-consciousness. As part of the project, INIT will use its expertise in public transport data structures and, together with its project partners, will process different mobility data using AI processes.

Customer-friendly, efficient on-demand transport from first to last mile

Our integrated booking, dispatching and optimization solution for on-demand transport MOBILE-FLEX is another example of the use of artificial intelligence. The topic of first/last mile transport plays a major role in the effort to attract more passengers to public transport as part of the mobility transition. For passengers, smooth transportation from the start of their journey, such as their home, to the stop where they board bus or train (first mile), and from the final stop (e.g. the main train station) to their actual destination (last mile) is becoming increasingly important. As a result, public transport companies are starting to recognize the importance of coordinated on-demand services, for example in rural areas or during off-peak periods.

With MOBILE-FLEX, we offer a product that supports all common forms of on-demand operation, from fixed-route services with individual on-demand stops, to corridor operations, to flex routing with ride pooling. Artificial intelligence also plays an important role, because an AI-based optimization algorithm efficiently and easily links passengers’ travel requests. The system’s excellent performance allows short booking deadlines based on real-time vehicle positions as well as ride pooling with virtual stops, addresses or geo-coordinates. By specifying fixed points and departure times (for the strategic positioning of vehicles, such as depots or stations), MOBILE-FLEX combines the flexibility of modern flex routing with the operational requirements of public transport.

Conclusion: Public transport is fast evolving. As mobility service providers for their region, public transport companies must use modern technology to enable this transformation in a way that benefits passengers. Machine learning and artificial intelligence are important tools to increase passenger satisfaction in order to deliver the mobility of the future right now.