Politecnico di Torino (logo)

In Hand Robotic Manipulation

Antonio Finelli

In Hand Robotic Manipulation.

Rel. Massimo Sorli. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Meccanica (Mechanical Engineering), 2022

PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (44MB) | Preview

In-Hand dexterity is the ability to make coordinated movements of the fingers to grasp and manipulate objects. For humans to handle an object is an innate ability and a straightforward task, but the situation is completely reversed if a robot is performing the manipulation. Traditional industrial robot manipulators are typically designed as the combination of an arm and a simple gripper as end-effector. Undoubtedly, this configuration is highly efficient when a high degree of on-site manipulation by the end-effector is not required. The challenge that fascinates researchers and robotics lovers is to attempt to develop end-effectors that could provide the most efficient manipulation feasibility. To achieve this purpose, multi-fingered robotic hands are employed. Despite robotics making giant strides every day, dexterous manipulation can be considered one of the most uncertain explored topics of robotics since requires a precision and an accuracy in the control of the variables involved that verge on perfection. The main objective of the thesis is to contribute to the research and development of robotic dexterous manipulation. In this regard, the question behind the project is the following: is it possible to perform manipulation by executing rotation of the grasped object around any axis in space? Starting from this request, a manipulation strategy is implemented, built on the idea that the kinematic of the fingers is used to change the object from an initial to a final configuration while maintaining fingertip contacts. The robotic device employed is the Allegro Hand, an anthropomorphic robotic hand made up of four fingers with sixteen degrees of freedom. The manipulated objects are rigid bodies, so it is assumed that, regardless of the forces applied, they do not experience any deformation. Moreover, the shapes of the objects are unknown. In fact, the procedure is based only on the tactile information obtained by tactile sensors located at the fingertips and on the kinematic structure of the hand. No other external feedback sources are considered. In a first phase, the resulting movements from the strategy that the Allegro's fingers must execute are tested in a simulative environment without considering the grasped object. Then, the algorithm is experimented in real environment with the aid of the simulative environment to verify the forces involved in the manipulation and the points of contact between fingertips and the grasped object. The results indicate that the algorithm is accurate in specific axes of rotation. Against this background, future research may be aimed at implementing a hybrid controller so that the optimal combination of force and position could be established. Additionally, more detailed research on the frictional forces involved in manipulation might lead to satisfactory results.

Relators: Massimo Sorli
Academic year: 2021/22
Publication type: Electronic
Number of Pages: 113
Corso di laurea: Corso di laurea magistrale in Ingegneria Meccanica (Mechanical Engineering)
Classe di laurea: New organization > Master science > LM-33 - MECHANICAL ENGINEERING
Ente in cotutela: IOC - Institute of Industrial and Control Engineering (UPC - Universitat Politècnica de Catalunya) (SPAGNA)
Aziende collaboratrici: Universitat Politècnica de Catalunya
URI: http://webthesis.biblio.polito.it/id/eprint/22399
Modify record (reserved for operators) Modify record (reserved for operators)