PhD Position Higher studies in Reinforcement Learning Algorithms, Robotics

    PhD Position Higher studies in Reinforcement Learning Algorithms, Robotics

    PhD Position for Graduates in Safety, Reinforcement Learning Algorithms, Robotics is available in the Department of Computer Science, IT University of Copenhagen, Denmark (Reinforcement Learning Projects)

    Deadline to Apply

    Apr 13, 2021 (23: 59 GMT +1)


    PositionPhD Position
    No. of Position(s)One
    Research Area– Computer Science
    – Reinforcement Learning
    ScholarshipAccording to Standard Norms
    WorkplaceDepartment of Computer Science
    IT University of Copenhagen
    Contract Period3 Years


    The following qualifications are required:

    • You have a genuine interest in the topics described above and a desire to work in an interdisciplinary environment.
    • You hold a computer science degree, or a similar, with good results in relevant subjects. Candidates graduating before September 1, 2021 are also welcomed to apply and should mention their expected graduation date.
    • You have background in methods for program validation (for example model-checking, program verification, program logics, abstract interpretation, or symbolic execution, automated testing) or
    • You have learned foundations of reinforcement learning (tabular and approximate methods). 
    • It is not required to have expertise both in formal methods and in reinforcement learning areas prior to enrollment. One of these is sufficient but knowing both will be seen as an advantage. You must be interested in and willing to work in both domains in the project. 
    • You have a critical, analytical, and creative mind, and enjoy problem solving.
    • You can take responsibility and meet deadlines prioritizing where necessary.
    • You have excellent written and verbal communication skills in English.


    You will work on safety and explainability of Reinforcement Learning using methods for software safety and reliability:

    • Reinforcement Learning has been a cornerstone of many success stories in modern AI and robotics.  It allows to learn agents that perform extremely well in computer games (Alpha Go, Dota2 games) and agents that perform tasks in robotics (for instance object grasping). At the same time, the correctness and reliability of the learnt policies remain very difficult to understand and assure, especially when learning happens online.
    • The aim of this project is to develop methods and tools that will enable industry to automatically learn correct and near-optimal controllers for safety-critical systems within a variety of domains. The challenge is to offer the advantages of AI based system (robustness to small changes, near-optimality, efficient learning of complex models) to the safety critical industry which requires strong guarantees about the behavior of the controller. 
    • The key research problem is to explore the design space between the controller synthesis (correct-by-construction) and reinforcement learning (efficient but based on sampling) to design learning methods that allow to qualify reliability of the resulting controller.

    How to Apply?

    To apply, please click APPLY ONLINE button mentioned above.

    Documents Required

    The application must be in English and must include:

    • A CV in English
    • Motivation letter / Statement of purpose.  Explain how your profile matches the position applied for.  What interests you in the area? List your most relevant skills for the position in bullet points. Maximum 1 page.
    • A list of past research experience and publications if any.
    • Copy of the degree diploma, with translation to English (if the degree qualification is not in English).
    • Transcript of your grades from the qualifying degree (along with an English translation if needed). We strongly encourage including the transcript from undergraduate studies as well.
    • Proof of English proficiency. You can use a TOEFL certificate, another exam certificate at similar level or higher, a high school diploma showing English results from several semesters, graduation diploma from an education you took in English, a substantial single author publication in English, Master’s thesis, other substantial text authored in English, etc.
    • Optionally up to two recommendation letters from academics or industrial experts in relevant areas or contacts to at most two reference persons.

    About the Project

    The PhD project will be run in collaboration with Aalborg University (prof. Kim Guldstrand Larsen), with University of Oslo (prof. Einar Broch Johnsen), and with industrial partners: HOFOR, Grundfos, Seluxit, and Aarhus Vand.

    This PhD project is part of the Danish Centre for Digital Technologies (DIREC), whose vision is to become an international leading centre with respect to research and education in computer science. DIREC is an alliance between the strongest computer science departments in Denmark bringing together Danish research strongholds and leading researchers across departments. The PhD student will also be affiliated with the REMARO Marie Curie training network on Reliable AI that is coordinated by the same research group. REMARO gives access to a vibrant research environment on relevant topic, and provides access to extensive training in safety, reliability, AI and robotics that can support the PhD student.

    You will be a part of a lively interdisciplinary Software Quality Research Group (SQUARE) at IT University.  The topics studied in SQUARE range from software quality issues ranging from safety and security, verification and program analysis, through testing, modeling, language design, architecture, visualization, and human and social aspects.



    Professor, Andrzej Wasowski
    IT University of Copenhagen

    Official advertisement

    PhD Position in Formal Methods for Safety Reliability and Explainability of Reinforcement Learning

    Other PhD Positions with Scholarships

    Post expires at 8:59am on Wednesday April 14th, 2021 (GMT+9)