Traveling Bus Problem - The Optimization Problem on a Bus Depot

Context

Digitalization and the trend toward automating driving movements make it possible to reduce costs in the local public transport sector and additionally increase safety on the road. For this purpose, depots, as the heart of public transport vehicle fleets, offer themselves as starting points. On the one hand, a depot represents a closed, controllable area, which is why it is suitable as a field for testing vehicles of SAE level 4 (and higher). Second, the StVO applies within this area, making it a springboard for transferring automated driving movements to the public domain.

Content

In depots, vehicles accrue several missions - at different stations - on the premises. These missions vary by vehicle type, but also daily per vehicle. Therefore, the optimal sequence is to be recalculated ad hoc for each vehicle when it enters the depot. This calculation takes place centrally and is subsequently issued to the individual vehicle as a series of missions. An optimum of the route to be taken and a coordination with the station occupancy offer enormous potential for time savings.

Objectives

• Literature review on approaches of classical, agent-based and machine learning based routing algorithms.
• Evaluation/comparison of representatives of the three mentioned approaches with respect to different criteria (available computing capacity, runtime, feasibility on the vehicle, etc.).
• Implementation of the algorithms and subsequent evaluation in terms of time savings with respect to the total travel distance of a single vehicle.

Requirements

• Programming knowledge Python
• Independent and solution-oriented way of working
• German or English business fluent in word and writing