• analysis
• force-closure
• grasping
• manipulation
• module
• open-source
• optimization
• package
• robotic hands
• toolbox
• whole--limb
• whole-arm

To afford the ever increasing level of power and flexibility demanded by sophisticated applications, robots must be able to interact with and directly effect change on highly unstructured environments. This interaction can take multiple forms; in this thesis, we focus our attention on two important forms of robot-environment interaction, namely the task of restraining objects, also called grasping, and the task of manipulating objects, sometimes called dexterous manipulation. We discuss the implementation of an high-level, open-source, modular simulation platform which serves as a bridge between theoretical calculations and laboratory experiments, and which is built on top of Sci(cos)lab/Scicos and of its Robotics Toolbox (RTSS), developed at Centro~``E.Piaggio'' since 2007.

After a review of the main solutions currently available in the software field of robotic manipulation, we motivate our choice of improving and extending RTSS capabilities to grasp and dexterous manipulation simulation. Grasping and dexterous manipulation of objects are studied within the cooperative manipulation paradigm, where an arbitrary number of robot limbs, simple chains of links connected through revolute or prismatic joints, are coordinated towards task accomplishment; the kinematic and quasi-static models of the grasp are reviewed, and the most relevant results about their analysis are reported. Then, the analytical frameworks are reconsidered in terms of object-oriented development, and transformed into software code. More specifically, we identify all the components of an abstract representation of a generic manipulation system, be it a cooperating multi-arm system, a robotic hand or a parallel structure robotic finger, and for each component we define the properties (members), describe a collection of actions associated with it (methods), and design a public interface allowing valid changes in its data (data encapsulation). The effectiveness of the proposed approach and the correctness of the implemented functions are then validated by assessing numerical results of the simulation of several case studies, reproduced from some of the most relevant references on the subject of grasping and dexterous manipulation.