Vision-guided feedback control of a mobile robot with compressive measurements and side information

In this work, we present feedback control laws for vision guided navigation of a mobile robot. The robot is modeled as a cart that can move along a straight line, and has two vision sensors onboard. The primary vision sensor is a high resolution single-pixel camera (SPC), based on principles of compressive sensing, for capturing images. Additionally, there is a low-resolution sensor that provides coarse measurements. In this work, we consider a simple scenario in which the target is modeled as a straight line segment on a plane. The main contribution of this work is the formulation of control laws directly from the compressed measurements, obtained from the SPC. Therefore, the reconstruction of the target image is sidestepped, leading to a reduction in the amount of data acquired for control.