Unique features of MOTIS

MOTIS is a timetable and mobility information system.

You can find an introductory presentation of the Department of Algorithmics at the Darmstadt University of Technology at https://www.algo.informatik.tu-darmstadt.de/#motis (German).

The following chapters point out all the unique features which, as far as we know, are not found in other systems.

Mathematical optimality is guaranteed

The basics of all commercial train and transport association systems on the market were founded in the 1980s. For those who remember, it was the time of the Intel 286 processor and the arrival to the market of the 386 -  the first processor for personal computer systems able to support real operating systems. Optimizing algorithms like a timetable information system faced massive problems with long and unacceptable computing times. Since computing times rise disproportionately with increases in the amount of data, developers had to be creative. Thus, they invented heuristics to control computing time. The quality was still good, but these solutions did not necessarily deliver optimal results.

When we tried to implement a commercial system for a large German public transport service provider, we found that setting all the parameters to control the heuristics became a puzzle that was too difficult to solve. This was particularly the case with regard to allowing for further functional development or large extensions of the timetable data.  The reason for this was partly because of the complexities resulting from side effects between different parameters which would not allow simple changes, such as improving quality at the expense of reducing performance.. During the process of making a large functional extension of the timetable information system for this particular customer, we came up with the idea of developing another timetable information system based upon the principle of diverse redundancy. We chose this approach because the optimality of the results of a complex system like a timetable information system based on heuristics cannot be proven. The new system aimed to show the potential for improvement of the commercial system.

One of the main targets therefore was to guarantee mathematical optimality without relying on the use of heuristics. Not even at the start of the  21st century was it possible to allow travellers to get an optimal result from an information query except at an unacceptable cost of computing. But this is not an issue which is relevant in the context of a quality assurance system. Today, Moore’s law remains as valid as ever. This means that mathematical optimality still requires comparably as much computing time but, in a commercial system, it is not really relevant any more since the price-performance ratio has changed drastically.

Regarding MOTIS, we have done lots of research on acceleration techniques and have also tested heuristics with good results. MOTIS’s advantage, however, is that the quality of such an accelerated MOTIS could be tested by a non-accelerated MOTIS which guarantees optimality. This would mean that diverse redundancy was realized by MOTIS itself.

Multi-criteria search

During the mid-1990s, commercial timetable information systems were still designed to deliver only the fastest routes. But, with the spread of electronic timetable information systems as the base of ticketing for transport providers and travel agencies, it soon became obvious that only making the singe form of optimization does not satisfy the customers’ needs.

Except for offering the fastest routes, it was not possible to offer alternatives which, for instance, had less train changes with only a few more minutes of travel time. The routes offered were unnecessarily complicated, especially for disabled people, travellers with small children, or those carrying large baggage. The desire arose for “convenient” travel alternatives involving less train changes with not much longer travel time. And, of course, the desire for “inexpensive” alternatives was always present. Therefore, in ticket sales offices, either the specialised knowledge of the salesperson was used or simple “trial and error” was applied in order to find the least expensive travel alternative.. Classically, the enquiry was entered with the parameter “without ICE / IC” (ICE and IC are German high-speed train categories) in order to find less expensive connections with acceptably longer travel times. However, travellers may have other criteria to choose from as a result of their enquiry. How crowded is the train? Does it have air-conditioning? Does it have WiFi? Is there a restaurant / bistro? Can I reserve seats? Are there recreation areas? How reliable are my routes? Is the landscape interesting which I am going to travel through? Which journey has the least CO2 emissions?

For this reason, MOTIS, by its very nature, is designed to consider any number of search criteria. If we imagine each travel alternative as a vector representing several criteria, then, for instance, we can represent an itinerary as a vector with the following components: travel time, number of train changes and prices. For the computation of deliverable alternatives, the basic principle that applies is that one alternative will only repress another if it is better in one criterion but not worse for any other criterion. This basic principle has to be honed, of course, to ensure that the time of day and length of journey are within reasonable limits which are acceptable to the traveller.. For example, nobody would accept an extension of travel time from 2 to 3 hours only to avoid one change of train. Thus, it’s important to realise that the basic principle of MOTIS is its ability to include multiple criteria in a simple way.  

A practical example was revealed by experiments to improve the night train information, which was still not very good at the beginning of the 21^st century. This was because night trains, as well as ferry boats, are much slower than regular trains. The result was that commercial systems could seldom find night trains or ferry boats due to the underlying basic principle that finds the fastest routes.

The reason for the demand for night trains and e.g. Baltic Sea ferry boats is often that the traveller is able to sleep uninterruptedly for several hours on board. As a result, travel time loses its importance. For MOTIS, therefore, a module called “night train information” has been developed in which alternative itineraries including uninterrupted sleep time are not excluded,  as long as at least 6 hours of sleep is possible. The option of including a requirement for sleep thus became a new search criterion.

A similar solution could be found in relation to a form of travel information within the whole of Europe that includes flight data. Since the aeroplane is considered faster than the train for distances greater than 600 km, trains would mostly be excluded for such distances. However, by applying appropriate rules, MOTIS could make it possible to automatically replace train journeys only if travelling by aeroplane was faster by at least one hour for the same distance. Therefore, it would still be possible to have an optimal combination of land-based and air-based traffic through a multi-criteria search. This represents a functionality that guarantees optimality according to given criteria – something that we still do not have even today in commercial information systems. 

Intermodality with individual transport

Since 2013, MOTIS is being continually extended to support any other mode of transport (e.g. bicycle, car, taxi, long-distance walks, etc) in addition to the public timetable-dependent transport. The integrated algorithmic approach used to calculate all journey in the same network,  which comprises the whole data pool rather than a distributed system, guarantees the optimality of journeys with adequate enquiry response time.

Many information systems only include walking within a “search from door to door” which has a bus or train stop as a part of the journey.

MOTIS overcomes these spatial limitations and allows, for instance, the inclusion of bicycles for distances of 5 km or park & ride parking for journeys of more than 10 km with a private car, especially in rural areas. This represents a real improvement in the information available for intermodal travel chains. It also supports “kiss and ride” where, for example, the commuter is taken to a train, tram or bus station because there is a lack of public transport for the "last mile" at off-peak times, which may be combined with an off-peak option(e.g. during the weekend, early in the morning or late in the evening).

Intermodality with Ride Sharing

In 2014, in cooperation with flinc (German) and Deutsche Bahn, MOTIS was extended as a prototype in order to find dynamic ride hailing offers in addition to public and private transport offers. Dynamic ride hailing offers can be used from or to a train stations, but also for distances between two train stations. The very promising results of the proof of concept could not be further developed because flinc was taken over by Daimler AG in September 2017. The public platform was closed by the end of the same year. According to our knowledge, this is the first integrated approach that combines timetable-dependent public traffic with ride sharing offers.

Intermodality with transport on demand

Initial considerations appeared to suggest that solutions like those mentioned above would also be suitable for (autonomous) transport on demand. The advantage of MOTIS’ integrated approach, in comparison with other approaches, is that this kind of transport can be combined immediately with any other kind of transport within travel chains. Thereby, travel time of trips which are independent of a timetable are based on empirical values such as navigation systems for vehicular traffic. This integrated approach allows for reliable judgements about the quality of resulting travel chains, which have clearly lower response times when compared with distributed (networked) systems. 

Further MOTIS oriented research

The following dissertations are currently underway in varying degrees of completeness in the Department of Algorithmics within the Faculty of Informatics, led by Prof. Dr. Karsten Weihe:

• Intermodal real-time search for connections
• Forecast of traveller flows (in cooperation with Deutsche Bahn long-distance traffic)
• Optimal control of traveller flows (strategies in the case of obstacles)

MOTIS looks back on a long history of research and has thereby developed many features which are even today still to be considered innovative. We would like to work with partners to further develop these features up to a productive system.

We would also like to work on research projects with interested partners to develop actual research approaches, particularly in the context of combining “intermodal” with timetable-dependent modes of transport with on-demand offerings.