Department of Computer Science Technische Universitaet Muenchen
Home People Teaching Research Publications Info Sitemap
Research | Publications | Teaching | Miscellaneous | Contact

Research - Freek Stulp

My main research objective is to implement control approaches for autonomous robots that will lead their behavior to be as robust, versatile, and adaptive as our own. To do so, I believe we must understand and exploit the general principles underlying animal and human motor control. For instance, my PhD research was inspired by human forward models, and focussed on enabling robots to autonomously learn similar action models. In my thesis and related publications, I demonstrate how these action models enable a variety of robots to optimize and coordinate their behavior. Robots should be designed to help people. An example is robots that assist and support the elderly, as in the SHARE-it project. Note that human-like robustness, versatility and adaptivity will be essential properties of robots if they are to assist people in unstructured environments.
Apart from this primary research theme, I also have an interest in Image Understanding, and Knowledge-based Systems.

Here is a list of research projects I was/am involved in. Years refer to my involvement, rather than the duration of the project.

  • CogMan [2008-]
    In the CogMan ("Cognitive Models of Everyday Manipulation Tasks") project, we develop a computational and control model of pick-and-place tasks in the context of everyday manipulation activities in human environments. This is done based on the results of tracking the gaze and pose of humans during manipulation tasks. These models are then implemented into a control system for a articulated robot. We empirically analyze the impact of this control model on the flexibility, robustness, adaptability, and naturality of the robot behavior. CogMan is part of the CoTeSys cluster of excellence.

  • SHARE-it [2007]
    The aim of the EU-funded Share-it project is to provide assistive technology for the elderly, so that they can live at home as long as possible with a maximum safety and comfort; thus increasing their quality of life. This technology consists of semi-autonomous wheelchairs, biometric monitoring and ambient intelligence. My role within this project is to manage and coordinate the integration and evaluation of these technologies at hospitals, home care institutions and Ambient Assisted Living environments. Please also see my publications on ambient assisted living for more information.

  • AGILO: Semi-automatic Acquisition of Visuo-motoric Plans [2002-2006]
    My PhD (defended 10/2007) research is part of this project. The research focuses on providing robots with forward models, that are able to predict the outcome and performance of actions. The robots acquire this knowledge by learning forward models from observed experience. On-line, these forward models can be used to tailor actions to the current task context.
    One of the robotic platforms on which I implemented these ideas were the Pioneer I robots of our RoboCup mid-size league team "The AGILO Robocuppers". Media coverage on this team can be found here.
    Please also see my publications on motor control or robotic soccer for more information.

  • Learning Objective Functions for Face Model Fitting [2006- ]
    Together with Matthias Wimmer, I am researching the use of learning algorithms to acquire objective functions, one of the three main components in model fitting. These objective functions are learned using the same algorithms used for the robotic forward models described above. There is an on-line demo of the algorithm. Publications on this topic can be found here.

  • CAMERA [2000-2001]
    In the EC funded TMR network CAMERA, I have done research on describing range images with surface models, to detect and complete occlusions. A summary of the work can be found here. Publications on this topic can be found here.

  • KB-TCP [1999-2002]
    As part of a student project, I did research on visualizing a knowledge-based version of the Transmission Control Protocol (TCP), and verifying it with epistemic logic. Relevant publications can be found here.


Computer Science IX | Technische Universität München