Main menu:
Student Projects
Evolutionary Multi objective Optimization (EMOO)
The computational models developed at the CPL require intensive parameter tweaking before they run successfully. At present this parameter optimization is done by hand. However, Evolutionary Multi objective Optimization (EMOO) offers a possibility to implement some degree of automation for this process. The task is to implement a variety of EMOO-approaches and test their suitability for the computational models.
Five-colouring algorithm for perceptual grouping
It is clear from experimental research in cognitive psychology that perceptual grouping processes and visual attention interact with each other. The aim of this internship is to equip a computational model of visual attention (SAIM; Selective Attention for Identification model, http://www.comp-psych.bham.ac.uk/saim.php) with a grouping process. There has been a first attempt ( http://www.comp-psych.bham.ac.uk/publications/wapcv2004.pdf), but this version of SAIM works only under limited conditions. Recently it has become clear that the five color map theorem ( http://en.wikipedia.org/wiki/Four_color_theorem) can form the basis for a more promising approach to perceptual grouping in SAIM. In a first step your task is to implement and test the five-colouring algorithm in Matlab. The second step is to design and implement a biological plausible version of this algorithm in C++. In a final step this neural network has to be integrated into SAIM.
Control of a robot arm
The CPL has recently acquired a simple robot arm. The objective of this project is to simulate human visually guided arm movements in a multiple object scenario. A stereo camera system will provide the necessary visual information. The aim of this internship is to contribute to this overall aim.
Possible tasks are:
Generation of smooth trajectories for given start and end points.of object positions from stereo camera pictures. of a speed control for DC servos.
Interactive software tool for teaching inferential statistics
Teaching inferential statistics is often challenging, partially because inferential statistics is based on complex mathematical concepts and partially because the mathematical skills of the learner are not sufficient for these mathematical concepts. In order to improve teaching an ''interactive software tool for teaching inferential statistics'' was developed for the analysis of variance (ANOVA). The software tool allows learners to explore the relationship between the outcome of an ANOVA and properties of experimental data in an interactive fashion. Such ''learning by doing'' substantially improves the teaching of statistical methods. The aim of this internship is to debug the current version and extend it with additional statistical methods.
Automatic detection of grasp points
In a current experiment we take photos of how participants grasp objects in order to document the contact points of the fingers. At present the contact points have to be marked by the user. The aim of this work is to implement a computer vision program for an automatic detection of contact points.