In recent years, the rapid advancement of robotics has led to a surge of interest in the dexterous manipulation capabilities of robots, which holds the potential to address a wide range of practical problems in domains such as industrial manufacturing and household services. Consequently, numerous open and challenging research questions have emerged. This thesis presents a comprehensive investigation aimed at enhancing robotic manipulation abilities through the integration of perception and mechatronic system design, with a focus on real-world robot systems.

In the realm of robotic perception, contemporary robots are equipped with an array of sophisticated sensory devices, including depth cameras, and tactile sensors. This thesis capitalizes on the unique properties of these sensor typ...[ Read more ]

In recent years, the rapid advancement of robotics has led to a surge of interest in the dexterous manipulation capabilities of robots, which holds the potential to address a wide range of practical problems in domains such as industrial manufacturing and household services. Consequently, numerous open and challenging research questions have emerged. This thesis presents a comprehensive investigation aimed at enhancing robotic manipulation abilities through the integration of perception and mechatronic system design, with a focus on real-world robot systems.

In the realm of robotic perception, contemporary robots are equipped with an array of sophisticated sensory devices, including depth cameras, and tactile sensors. This thesis capitalizes on the unique properties of these sensor types to develop a series of perception systems tailored to objects with distinct physical attributes. For rigid objects, a perceptive robotic assembly system is proposed, which employs an in-hand RGB-D camera to facilitate peg-in-hole tasks involving various peg designs. Regarding deformable objects, a perception system is introduced that combines multi-modal visual and tactile information to accurately discern the state of textile objects during insertion tasks.

In terms of mechatronic design, this thesis presents the development of a lightweight modular robotic arm, which offers improved dynamic response and torque control capabilities. Furthermore, a novel end-effector is designed to facilitate the automated grasping of textile objects more effectively. Additionally, several iterations of lightweight mobile manipulators are developed to expand the manipulation workspace of the manipulator for further study.

The efficacy of the proposed algorithms is rigorously evaluated on real robot systems, while the feasibility of the mechatronic system designs is demonstrated through the successful completion of prototype development. This comprehensive approach to enhancing robotic manipulation capabilities through perception and mechatronic system design holds significant promise for advancing the field of robotics and its practical applications.