An exemplary application of decision theory in robotics: autonomous calibration of depth cameras with non-overlapping fields of view on a mobile platform

Abstract

This thesis is about the creation of a decider, which is able to evaluate decision trees and utility tables. The decider enables the robot to decide by its own, which action should be performed in the actual situation. Afterwards the decider is used for the calibration process of time of flight cameras with non-overlapping field of views on a mobile platform. A framework is the main objective of this thesis, which can be utilized for known problems as Localization and Exploration in mobile robotics. Therefore, clear interfaces will be designed, to ensure that the framework will be used in future works. After the implementation of the decider a decision tree is designed containing the calibration’s process. This tree must have clearly specified actions and states, which can be observed by the robot itself. At the end the combination of the decision process with the calibration will be evaluated.