paper supplements & projects
This page contains some supplementary material to papers or general projects. Some of these might be quite old. Please, refer to the papers for more details. (Use the full screen mode (lower right button) for better video quality.) ICRA 2009 submission

Marc Toussaint, Nils Plath, Tobias Lang, Nikolay Jetchev: Integrated
motor control, planning, grasping and highlevel reasoning in a blocks
world using probabilistic inference (submitted to ICRA 2010).
Here is the attached movie:
Get the Flash Player to see this player.  HUMANOIDS (2008) paper

Michael Gienger, Marc Toussaint, Nikolay Jetchev, Achim
Bendig, Christian Goerick: Optimization of fluent approach and
grasp motions.
8th IEEERAS
International Conference on Humanoid Robots (Humanoids
2008)
Here is the attached movie:
Get the Flash Player to see this player.  HUMANOIDS (2007) paper

M Toussaint, M Gienger, Ch Goerick (2007):
Optimization of sequential attractorbased movement for
compact.
(preprint)
Humanoids 2007.
Here is the attached movie:
Get the Flash Player to see this player.  Visual flow estimation & segmentation using belief propagation

M Toussaint, V Willert, J Eggert, E Körner
(2007): Motion Segmentation Using Inference in Dynamic Bayesian
Networks.
British Machine Vision
Conference (BMVC 2007).
best paper award: V Willert, M Toussaint, J Eggert, E Körner: Uncertainty Optimization for Robust Dynamic Optical Flow Estimation. The sixth Int Conf on Machine Learning and Applications (ICMLA 2007), pages 450457.
Get the Flash Player to see this player. Get the Flash Player to see this player. Get the Flash Player to see this player.  Probabilistic inference planning

Here are some movies: (red=forward messages, green=bwd messages, blue=posterior state visiting probability for random starts/goals)
Get the Flash Player to see this player.Get the Flash Player to see this player. Please refer to:
M Toussaint, S Harmeling, A Storkey (2006): Probabilistic inference for solving (PO)MDPs. Research Report EDIINFRR0934, University of Edinburgh, School of Informatics.
M Toussaint and A Storkey (2006): Probabilistic inference for solving discrete and continuous state Markov Decision Processes. 23nd International Conference on Machine Learning (ICML 2006)
M Toussaint, Ch Goerick (2007): Probabilistic inference for structured planning in robotics. Int Conf on Intelligent Robots and Systems (IROS 2007).
 Inference and planning on factor graphs
 See here for a currently ongoing project on implementing generic inference techniques (based on message passing) on factor graphs and using this for solving (PO)MPDs.
 Motor control and physical simulation
 We (mainly myself and Heiko from Edinburgh) are currently working on nice physical simulation environments. We use the Open Dynamic Engine as the physical simulation engine, but designed an additional generic data structure describing the dynamical state of the system for more sophisticated computations (inverse kinematics, etc). Here are some preliminary results: A movie demonstrating inverse kinematics for an arm (simply using the pseudo inverse) and a physically simulated 1legged hopper (Windows codec) that is quite stable and flexible (not a simple superposition of oscillation and stabalization). Source code will be published some time in the future.
 Bayesian Search & Gaussian Process priors

See the project page on Bayesian Search and Gaussian Process priors, which is mainly addressed to the optimization and EC community.  (2005) Learning discontinuities in inverse dynamics

Please see here as a supplement toM. Toussaint and S. Vijayakumar (2005): Learning discontinuities for switching between local models. 19th International Joint Conference on Artificial Intelligence (IJCAI 2005), 17441745.
 (2004) Sensorimotor maps

Please see here as a supplement to the publication
M. Toussaint (2006): A sensorimotor map: Modulating lateral interactions for anticipation and planning. Neural Computation 18, 11321155.
For earlier work on this, see here that refers to
M. Toussaint (2004): Learning a world model and planning with a selforganizing dynamic neural system. In Advances in Neural Information Processing Systems 16 (NIPS 2003), 929936, MIT Press, Cambridge.
 (2003) Lerning genetic representations

Please see here for a project on
evolving genetic representations:
M. Toussaint (2003): Demonstrating the Evolution of Complex Genetic Representations: An Evolution of Artificial Plants. Genetic and Evolutionary Computation Conference (GECCO 2003), 8697.