This master’s thesis presents a robotic system in which a manipulator mounted on a mobile platform acts as a measuring device for the platform’s pose. The pose information is fed back to the controller, enabling high-precision guidance of the platform during docking with a docking station and eliminating the need for additional sensors while exploiting the manipulator’s capabilities, otherwise intended for other tasks.
Chapter 1 details the docking problem of a mobile platform with a docking station and reviews current solutions in the literature, followed by the thesis objectives.
Chapter 2 introduces the hardware components of the robotic system, outlines their key characteristics, and explains their contribution to the control architecture.
Chapter 3 describes the software tools used for control and simulation of the robotic system.
Chapter 4 explains the camera operation and the algorithms employed for its calibration and for determining its pose relative to the connector on the charging station. The first algorithm relies on optical markers attached near the connector, whereas the second estimates the pose directly with a neural-network–based detector.
Chapter 5 describes the control strategy for the mobile platform in the various phases before docking with the docking station.
In the final Chapter 6 we first analyse the pose-estimation uncertainty of the different vision methods, then evaluate the reliability of manipulator docking with the station’s connector, followed by the overall reliability of platform docking. The chapter concludes with a discussion of the advantages and limitations of the proposed robotic system and suggests directions for future improvements.
|
