Postdoctoral researcher in prediction and optimization of railway operations

Despite the excellent quality of public transport systems in Switzerland, the railway system needs to increase its performance (quality, for instance, travel time) and capacity (amount of services run) and attract more travellers to match the ambitious targets from policy and environmental goals. One key aspect of traditional railway transport systems is their plan, which is based on predetermined routes, lines, and scheduled times, with little possibility of adjusting to unplanned and unexpected circumstances. The project focuses on optimal planning and rescheduling of railway operations.

To do so, large-scale optimisation, typically MILP (Mixed Integer Linear Programming) approaches are used, but they scale poorly and are somehow not directly applicable in industrial practice. To approach those gaps, it is crucial to learn the real-life constraints, and exploiting advanced computational methods (including pre-computation and/or machine learning) to make the solution process faster and consistently optimal. This position is to address this gap, together with the two largest industrial players in this field in Switzerland.

This position will be available, with an ideal starting date between February 2024 and April 2024, or further upon agreement; the planned duration of the initial contract is one year, to be extended based on successful performance, for up to 2 years.

The focus of this project is to understand how to find and generate solutions better and faster for the problem of temporal planning of railways. This challenge involves timetabling and rescheduling problems, potentially also considering the rerouting of trains on the network. For those problems, standard mathematical models exist, which combine optimization, with prediction of future operations. The research needs to

1) extend the understanding of the domain-related objectives, to grasp the complexity of the real-life decisions by human controllers, which consider future operations in their choices;

2) improve the solution of large-scale scheduling and routing problems by novel algorithmic, pre-computation, and/or machine learning approaches, which are used in decision support systems.

3) to quantitatively understand the usefulness of the suggested decision compared to the actions actually implemented in real life, to tune and improve the solution process.

The ultimate objective is to have a better, faster decision; and precisely understand what “better” means and how it can be mathematically modelled.

The project is designed in close collaboration with industry players operating as providers of control systems or as railway infrastructure managers. The project expects interaction with their existing systems, considering their ultimate customers, i.e. the transport operators. Integration in the optimisation team of the industrial partners is envisaged, which can also include part-time work at their site.

You ideally have (or you are about to receive) a Doctoral Degree in transport sciences, management/ decision sciences, econometrics, statistics, mathematics, computer science, physics or related fields. Your research track is consistent and shows a track record, or clear potential, for modelling, control, and optimisation of transport systems. You are highly motivated and self-driven, with a clear research vision, academic ambition, and excellent communication and writing skills (fluent spoken and written English is mandatory). Moreover, the following skills are expected of a promising candidate:

• Computer science and ability to program independently complex software
• Ability to model and work with algorithmic design, existing software and data provided by industrial partners
• Knowledge of mathematical optimisation (MILP, IP, LP) and/or control sciences
• Team working and communication skills
• Knowledge of German or similar languages is not required but is a plus

You enjoy working in an interactive international environment with doctoral students, post-docs and senior scientists, referring continuously to practical problems and solutions. 

ETH Zurich is a family-friendly employer with excellent working conditions. You can look forward to an exciting working environment, cultural diversity and attractive offers and benefits.