If you are looking for a bachelor/master thesis or a job as a student research assistant, you may find some interesting opportunities on this page.

In the field of Machine Learning, numerical optimization techniques play a focal role. However, as models grow larger, traditional implementations on single-core CPUs suffer from sequential execution causing a severe slow-down. In this thesis, state-of-the-art GPU frameworks (e.g. CUDA) are to be investigated in order implement numerical optimizers that substantially profit from parallel execution.

Requirements:

• Skills in numerical optimization algorithms
• Good programming skills in Python and C/C++

Contact: Daniel Nyga

Markov Logic Networks (MLNs) combine the expressive power of first-order logic and probabilistic graphical models. In the past, they have been successfully applied to the problem of semantically interpreting and completing natural-language instructions from the web. State-of-the-art learning techniques mostly operate in batch mode, i.e. all training instances need to be known in the beginning of the learning process. In context of this thesis, online learning methods for MLNs are to be investigated, which allow incremental learning, when new examples come in one-by-one.

Requirements:

• Experience in Machine Learning.
• Experience with statistical relational learning (e.g. MLNs) is helpful.
• Good programming skills in Python.

Contact: Daniel Nyga

In the context of the European research project RoboHow.Cog [1,2] we are investigating methods for combining multimodal sources of knowledge (e.g. video, natural-language recipes or computer games), in order to enable mobile robots to autonomously acquire new high level skills like cooking meals or straightening up rooms.

The Institute for Artificial Intelligence is hiring a student researcher for the development and the integration of probabilistic methods in AI, which enable intelligent robots to understand, interpret and execute natural-language instructions from recipes from the World Wide Web.

This HiWi-Position can serve as a starting point for future Bachelor's, Master's or Diploma Theses.

Tasks:

• Implementation of an interface to the Robot Operating System (ROS).
• Linkage of the knowledge base to the executive of the robot.
• Support for the scientific staff in extending and integrating components onto the robot platform PR2.

Requirements:

• Studies in Computer Science (Bachelor's, Master's or Diploma)
• Basic skills in Artificial Intelligence
• Optional: basic skills in Probability Theory
• Optional: basic skills in Machine Learning
• Good programming skills in Python and Java

Hours: 10-20 h/week

Contact: Daniel Nyga

[1] www.robohow.eu
[2] http://www.youtube.com/watch?v=0eIryyzlRwA

Developing new activities and improving the current simulation framework done under the Gazebo robotic simulator. Creating a custom GUI for the game, in order to launch new scenarios, save logs etc.

Requirements:

• Good programming skills in C/C++
• Basic physics/rendering engine knowledge
• Gazebo simulator basic tutorials

Contact: Andrei Haidu

Integrating the eye tracker in the Gazebo based Kitchen Activity Games framework and logging the gaze of the user during the gameplay. From the information typical activities should be inferred.

Requirements:

• Good programming skills in C/C++
• Gazebo simulator basic tutorials

Contact: Andrei Haidu

Improving the skeletal tracking offered by the Leap Motion SDK, by using two devices (one tracking vertically the other horizontally) and switching between them to the one that has the best current view of the hand.

The tracked hand can then be used as input for the Kitchen Activity Games framework.

Requirements:

• Good programming skills in C/C++

Contact: Andrei Haidu

Gazebo currently only supports rigid body physics engines (ODE, Bullet etc.), however in some cases fluids are preferred in order to simulate as realistically as possible the given environment.

Currently there is an experimental version of fluids in Gazebo, using the Fluidix library to run the fluids computation on the GPU.

The computational method for the fluid simulation is SPH (Smoothed-particle Dynamics), however newer and better methods based on SPH are currently present and should be implemented (PCISPH/IISPH).

The interaction between the fluid and the rigid objects is a naive one, the forces and torques are applied only from the particle collisions (not taking into account pressure and other forces).

Another topic would be the visualization of the fluid, currently is done by rendering every particle. For the rendering engine OGRE is used.

Here is a video example of the current state of the fluid in Gazebo.

Requirements:

• Good programming skills in C/C++
• Interest in Fluid simulation
• Basic physics/rendering engine knowledge
• Gazebo simulator and Fluidix basic tutorials

Contact: Andrei Haidu

Computer vision is an important part of autonomous robots. For robots the image sensors are the main source of information of the surrounding world. Each camera is different, even if they are from the same production line. For computer vision, especially for robots manipulating their environment, it is important that the parameters for the cameras in use are well known. The calibration of a camera is a time consuming task, and the result depends highly on the chosen setup and the accuracy of the operator.

The topic for this thesis is to develop an automated system for calibrating cameras, especially RGB-D cameras like the Kinect v2.

The system should:

• be independent of the camera type
• estimate intrinsic and extrinsic parameters
• calibrate depth images (case of RGB-D)
• integrate capabilities from Halcon [1]
• operate autonomously

Requirements:

• Good programming skills in Python and C/C++
• ROS, OpenCV

[1] http://www.halcon.de/

Contact: Alexis Maldonado and Thiemo Wiedemeyer

Create models during runtime for unknown textured objets based on depth and color information. Track the object and update the model with more detailed information, completing it's 3D model from multiple views improving redetection. Using the robots manipulator pick up the object and complete the model by viewing it from multiple viewpoints.

Requirements:

• Good programming skills in C/C++
• strong background in computer vision
• ROS, OpenCV, PCL

Contact: Thiemo Wiedemeyer

Create a simulation environment that represents the robots current belief state and can be updated frequently. Use off-screen rendering to investigate the affordances these objects possess, in order to support segmentation, detection and tracking of these in the real world.

Requirements:

• Good programming skills in C/C++
• strong background in computer vision
• Gazebo, OpenCV, PCL

Contact: Ferenc Balint-Benczedi

Objects in a human environment are usually found in challenging scenes. They can be stacked upon eachother, touching or occluding, can be found in drawers, cupboards, refrigerators and so on. A personal robot assistant in order to execute a task, needs to detect these objects and recognize them. In this thesis a multi-modal approach to interpreting cluttered scenes is going to be investigated, combining the results of multiple segmentation algorithms in order to come up with more reliable object hypotheses.

Requirements:

• Good programming skills in C/C++
• strong background in 3D vision
• basic knowledge of ROS, OpenCV, PCL

Contact: Ferenc Balint-Benczedi