As the state of the art of computing technology is advancing, computer-based machines are starting to perform new types of tasks in less structured and less predictable environments. Possible scenarios could include a mobile robot wondering in a natural environment or a virtual agent exploring a huge astronomical database looking for new type of celestial objects. My research goal is the creation of computer programs that allow these machines to succeed through an adequate mining of the information acquired by the agent.
The kinds of algorithms I work with are mainly based on probabilistic reasoning. Currently I am working in an adaptive visual system able to selectively combine and exchange different algorithms according to the quality of their information. The approach keep track of the posterior probability of the state of the system, this can give you information about how far are you pushing the limits of the algorithms. For example, if you are using an algorithm base on color information and the posterior probability is flat of with many peaks, it is likely that it will fail, so it is time to start new algorithms based on different cues such as depth or motion.