Dexterous Manipulation Planning for an Anthropomorphic Hand
In this paper, we present some results about dexterous manipulation planning with an anthropomorphic hand. The task is to drive the grasped object from a start to a goal configuration. The planning algorithm automatically computes the finger motion and th
- PDF / 570,059 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 97 Downloads / 200 Views
Univ Paris 06 - CNRS UMR 7222 - ISIR, France Institut Pprime - UPR 3346 CNRS - Universit´e de Poitiers - ENSMA, France
Abstract In this paper, we present some results about dexterous manipulation planning with an anthropomorphic hand. The task is to drive the grasped object from a start to a goal configuration. The planning algorithm automatically computes the finger motion and the required regrasping motion i.e. when and how to relocate the contacts on the object’s surface. The planner relies on a previously presented method but some extensions were added in order to make it applicable to the hand used in the experiments (a four-fingered hand with a human-inspired kinematics). Some experiments, conducted on the real platform, are presented.
1
Introduction
Many hand designs and mechanical realizations have been proposed in the literature such as the DLR hand (Butterfass et al., 2001), UB-hand (Palli et al., 2011) and the hand of the Shadow robot company. Most of them have been used to validate grasp planners, grasping force computation or control laws. Planning the motion of complex dexterous manipulation tasks has lead to less works and to few experiments. The first theoretic works dealing with dexterous manipulation planning (DMP) have presented description of the configuration space and the kinematic and geometric constraints of the problem (Li et al., 1989; Montana, 1995). Trinkle and Hunter (1991) proposed a solution to plan manipulation tasks with a two-fingered planar hand. The work of Yashima et al. (2003) considers all the possible contact modes (sliding, sliding with rolling, with spinning, etc.) between the fingertips and the object. A global planner explores the configuration space of the object via a rapidly-exploring random tree ”RRT” (Lavalle and Kuffner, 2000). From a current configuration of the system hand+object, the planner finds the motion to reach a new object configuration after choosing a set of contact modes at random and after integrating the inverse dynamic
V. Padois, P. Bidaud, O. Khatib (Eds.), Romansy 19 – Robot Design, Dynamics and Control, CISM International Centre for Mechanical Sciences, DOI 10.1007/978-3-7091-1379-0_30, © CISM, Udine 2013
242
J.-P. Saut et al.
model of the system. Xu et al. (2010) describe the problem with the hybrid automaton formalism. A RRT-like algorithm is used to solve the problem. Recently, Bouyarmane and Kheddar (2011) proposed a generalized framework to plan a discrete sequence of multi-contact stances that brings a set of collaborating robots and manipulated objects from a specified initial configuration to a desired goal through non-gaited acyclic contacts with their environment or among each other. The applications of DMP algorithms to real modern robotic hands are very few. Xue et al. (2008) focused on the particular task of rotating an object. The authors applied their technique on the Schunk anthropomorphic hand (Liu et al., 2008). Experimented were conducted to plan and execute the task of screwing a light bulb. However, the limited manipu
Data Loading...
