Recent developments in automated vehicle location (AVL) systems which provide the location of vehicles in real-time, and the use of smartphones with incorporated geolocation technologies, create a window of opportunity for new mobility services. As a result of these technologies, a shift from traditional car ownership and the use of traditional public transport to mobility-on-demand services can be observed. Examples of these services are car- or scooter-sharing (e.g., Poppy), bike-sharing (e.g., Velo), ride- or taxi-sharing (e.g., Uber, UberX Share), on-demand carpooling (e.g., BlaBlaCar) and even on-demand bus transport (e.g., Kutsuplus). These relatively new services are more flexible compared to traditional public transport lines. In addition, they are beneficial to counteract congestion in cities and increase the accessibility to large public transport hubs.
With these new services arise new optimization problems. When a user wants to take a bike from the train station to their office in the morning, chances are that the bike station is empty due to the morning-peak demand. As a result, bike-sharing operators should relocate their bikes over the city so that the number of bikes at each station is adapted to the demand. This gives rise to the Bike Request Scheduling Problem, which aims to determine the routes of the bike-repositioning vehicles.
Furthermore, as traditional public bus transport services are embracing the new technologies to better optimize their routes, both semi-flexible and fully-flexible alternatives arise. The first combines the positive characteristics of fixed-route and fully on-demand services by serving a mandatory set of bus stops added by a set of bus stops that can be requested on-demand. On the other hand, fully flexible on-demand bus systems build routes from scratch and are completely dependent on requests for transportation. In contrast to ride-sharing services, mini-buses are used and passengers are more heavily pooled. Both systems result in a continuous optimization of the bus routes.
Because mobility-on-demand services often need an online optimization approach, heuristics are preferred over exact solution algorithms.
- Vergeylen, N., Sörensen, K., & Vansteenwegen, P. (2020). Large neighborhood search for the bike request scheduling problem. International Transactions in Operational Research, 27(6), 2695-2714.
- Montenegro, B. D. G., Sörensen, K., & Vansteenwegen, P. (2021). A large neighborhood search algorithm to optimize a demand-responsive feeder service. Transportation Research Part C: Emerging Technologies, 127, 103102.
- Melis, L., & Sörensen, K. (2022). The static on‐demand bus routing problem: large neighborhood search for a dial‐a‐ride problem with bus station assignment. International Transactions in Operational Research, 29(3), 1417-1453.