PhD Position Autonomous Reasoning on Robots Operating in the Real, Open World

Interested in working on autonomous robots for the real, open world in a multidisciplinary team and also doing a PhD? Join us!

You will work within the OpenBots laboratory creating autonomous robots that operate in the real world with various other PhDs and engineers. The OpenBots laboratory aims to develop a common sense capability for robots making them able to understand their surroundings and plan proper actions to execute their inspection task, even when part of their surrounding world is unknown and dynamic. The inspection task deals with a robot walking around a perimeter and, meanwhile, detecting abnormal events based on vision or sound, after which a detailed investigation of the siutation might be required. The common sense capabililty that we develop shall allow the robot to complete similar inspection tasks in different (potentially changing) environments, such as governmental buildings, industrial plants, or train tracks. Our developments shall follow the latest achievements in artificial intelligence, world modelling, knowledge representations, mission planning and motion planning, where each team member has its own specialization.

Your research will be related to world modelling and knowledge representation, i.e., creating the robot's situational awareness, assuming that it is able to detect and observe individual objects and events through its camera and microphone. More specifically, you will study and develop algorithms for managing the robot's information based on these observations and based on a predefined, or learned, knowledge structure. Since our robots will operate in the real, open world, incoming information will be corrupted by noise and mistakes. Therefore, it is important that information is managed by combining evidences related to the observations with logic and symbolic reasoning, like a Bayesian network or Markov logic networks. However, to cope with the unknowns it might be essential to replace Bayes' principles with, for example, possiblity theory. In addition, we expect to develop deductive reasoning so that when the robot encounters an unknown situation it may rely on general knowledge/models, and plan experiments to either accept or falsify the application of such knowledge/models to the unknown situation. Finally, you will also study and develop algorithms that extract a situational graph representing the current situation of the robot, or its near-future. Situational graphs shall be used for task planning of the robot and should thus contain which task can be carried our where, for example through inductive reasoning on the available information along with patterns in such information that may be linked to a task-capability of the robot.

The OpenBots project is being carried out in a team of 5 PhD students, 3 at the Delft University of Technology (where this vacancy is one of such PhDs) and 2 at the University of Amsterdam. You will work 3 days a week at the university and the other 2 days you will work with all other 5 PhDs at one of the external partners being TNO and the Royal Netherlands Marechaussee where a physical lab environment is available. The project is carried out with supervisors from the Cognitive Robotics Group of the Delft University of Technology (Dr. J. Sijs, Dr. C. Hernandez Corbato and Prof. J. Alonso-Mora) and the Video and Image Sense Lab of the University of Amsterdam (Prof. Cees Snoek).      

Requirements

Need to haves:

• Graduated in one of these fields: System and Controls, Artificial Intelligence, Computer Science, Mechanical/Electrical Engineering or Applied Mathematics (or related).
• Interest in robotics, artificial intelligence and data science.
• Demonstrable experience in world modeling, machine learning, probabilistic reasoning and logic reasoning.
• Solid mathematics foundations, especially statistics, linear algebra and discrete mathematics (graphs, logic);
• Team player and willing to work in a multidisciplinary team of engineers, technicians and scientists.
• Programming skills using languages such as Python/C++, as well as robotic tools, such as ROS and Gazebo.
• You are highly motivated, independent, and creative.
• Good communication, presentation and writing skills and proficient command of English.

Nice to haves:

• Experience with publishing scientific work.
• Experience with embedded software and hardware.

Graduate Schools Admission Requirements.

will introduce you to some of our researchers and their work.