• Optimization
• control
• Forgetting factor
• active sensing
• kalman-like observer

The goal of this thesis work consists in the implementation of an optimal perception-aware control strategy for a mobile robot with a "forgetting factor" on past measurements which gives more weight on the most recent, and hence most updated, information data. The aim is to explore the possibilities offered by such a strategy in driving the robot along a trajectory that maximizes the information about the robot state in spite of intermittent measurements and the forgetting factor. To quantify the amount of collected information, the minimum eigenvalue or the trace of the Constructibility Gramian are considered as possible metrics taking also into account the effect of the forgetting factor on past measurements. To show the efficacy of the method in maximizing the estimation accuracy, I considered a case study involving a unicycle that measures the distance from some markers (whose position is known) and landmarks (whose position is unknown) located in a given area and needs to estimate its own configuration and the position of the landmarks.