PhD position on static and dynamic optimization for transport and logistics

Eindhoven University of Technology (TU/e), Netherlands

Updated: almost 2 years ago
Deadline: 29 May 2022

We are looking for a PhD candidate with a background in operations research, mathematics, econometrics, industrial engineering, or computer science. Especially candidates with the interest to work on the interface of optimization and AI are highly encouraged to apply. The project is centered around the development of new optimization methods (using machine learning, operations research, AI, deep reinforcement learning) needed for solving challenging operations management problems in, for example, retail operations, transportation, and logistics.

Project Description
Organizations frequently make decisions while facing an uncertain future. For example, inner-city stores are replenished by trucks before sales are known, and ICU beds are reserved each day before COVID-19 patient inflow is known. Making good decisions is essential for organizations and society. For example, congestion and pollution from trucks in inner-cities are reduced, and COVID-19 and regular patients receive better healthcare.

In this project, we focus such problems that require both a structural decision and daily dynamic decisions. The structural decision, for instance the reservation of ICU capacity for the coming days or weeks, has to consider the structure of the daily dynamic decisions, for instance the allocation of patients between hospitals.

Traditionally, (mixed) integer optimization is the method of choice for determining structural decisions, and when taking uncertainty into account methods are typically grounded in Stochastic Programming or Robust optimization. Alternatively, Markov decision processes and (nowadays) deep reinforcement learning/AI are the relevant fields for determining dynamic policies. How to combine both approaches is an exciting new research field in which this project will make fundamental and applied contributions.

The project will revolve around combining methods from both fields, developing new and novel solution approaches, and applying them on practical use-cases from retail operations and/or transportation. The project is envisioned to take place in three (related) steps:

  • Study the theory of joint structural and dynamic decision-making.
  • Using machine learning methods to learn a smart and compact representation of the dynamic decision problem that can be included in the structural decision problem.
  • Development of advanced methods tailored towards specific use-cases, for instance in the replenishment of retail stores or balancing capacity/containers in (transport) networks. With many connections to industry and business this will be aligned as fit with the project progress.

    • You, as a successful applicant, will perform the PhD project outlined above. The research will be concluded with a PhD thesis. You will be supervised by dr. Albert H. Schrotenboer. A small teaching load is part of the job.

    Academic and Research Environment
    You will be part of the Operations Planning, Accounting & Control group (OPAC). OPAC currently consists of 25 staff members, 10 postdocs and 45 PhD students. The faculty teaches and conducts research in the area of operations planning and control in manufacturing, maintenance services, logistics and supply chains. Research is generally quantitative in nature, while many of the researchers also engage in empirical research. The OPAC group is responsible within the university for all teaching in the areas of operations management, transportation, manufacturing operations, reliability and maintenance, and accounting and finance, both at undergraduate and graduate level. The OPAC group has close collaborations with the industry, which gives direct access to challenging operations management problems, new technologies, and empirical data.


    Similar Positions