AI-supported on-demand transport

The research project had to take into account the transport company’s economic interests, the service provided to passengers and the municipality infrastructure. Spiesen-Elversberg offered the ideal conditions for all three factors.
The municipality is densely populated, is served by three bus lines, but has no railway station. On-demand transportation should improve mobility around the town and provide residents with connections to central transfer stops. In order to win them over, more than 100 on demand stops were set up within walking distance (maximum 250 meters) of every location in the municipality. Two seven-seater minibuses from Neunkircher Verkehrs GmbH with child seats and a wheelchair space were available.

INIT provided the technical basis for the mobility platform with its MOBILE-FLEX solution for on demand services. At its heart is the artificial intelligence-based optimization algorithm Advanced Optimisation Core from INIT subsidiary inola. The optimizer calculated the start and end stops, departure and arrival times and intermediate stops of journeys based on journey requests for a fully flexible routing. It always tried to combine the routes of the individual journeys to enable ride pooling and thus find the most passenger-friendly and economical route.

In order for fixed-route and on-demand services to be offered and monitored together in a central operations control center in the future, on-demand services must be integrated into the Intermodal Transport Control System (ITCS). In the KIMonoS research project, MOBILE-ITCS was used as the operations control system, which already integrates other INIT solutions such as MOBILE-PLAN, MOBILE-FLEX and MOBILEstatistics. In the MOBILE-PLAN planning tool, separate operating areas, operating times on the individual days of the week and vehicle deployment were defined. The journey and route were calculated and emerged dynamically during operation based on passenger bookings. Dispatchers tracked the test operation with the current vehicle positions via the geographical information system (GIS) on a map display in the ITCS; control center staff also had an insight into the passenger bookings in MOBILE-FLEX. The statistics system MOBILEstatistics was not used, but would have made it possible to analyze the journeys in detail.

Passengers needed to use the KIMonoS app, provided by INIT’s subsidiary HanseCom, to book journeys. They simply entered their journey request, i.e. start, destination, desired departure time and number of passengers, and they were immediately shown a suggested journey with the most suitable departure and arrival times and the corresponding stops. After confirming the journey suggested by the system, they received a booking number on the app, which they then presented to the driver as authorization to travel when boarding the vehicle. As KIMonoS was running in a test mode, all journeys were in fact free of charge.

If several passengers booked journeys in the same direction, these were bundled together where possible (ride pooling). Even if the vehicles had already set off, journeys could still be added in real time – the optimizer took these bookings into account and rerouted the journey if necessary. 

Drivers also used an INIT application. There were no built-in on-board computers in the two vehicles used for the on-demand services. Instead, the COPILOTapp was used, which offers comprehensive on-board computer functions via an iOS or Android tablet. It took over data communication with the control center, sent location information to the ITCS in real time, and recognized departures and arrivals at stops. Driving personnel received new journey orders via the COPILOTapp using server interfaces and mobile communications. In addition to displaying the next stops to be approached, drivers could also be guided to the next stop using turn-by-turn navigation.

KIMonoS was INIT’s first on-demand public transport project with app support. It provided the perfect opportunity to test the integration of on-demand journeys with MOBILE-FLEX in MOBILE-ITCS in all its facets and to obtain
practical feedback. The routing algorithm was revised, and the stop positions were adapted with directional information so that vehicles always stopped at the correct place (and not on the wrong side of a busy road, for instance).

INIT’s on-demand solution MOBILE-FLEX is being developed with a focus on integration into public transport operations. This includes the use of autonomous vehicles, manufacturer independent integration with open standards and guaranteed connection protection for scheduled and on-demand journeys. After all, on-demand transport is becoming increasingly important, especially in rural areas.