Automatic motion of manipulator using sampling based motion planning algorithms - application in service robotics
|Other Titles:||Automatische Bewegung eines Robotarmes mittels Probenahme basierten Bewegungsplannung Algorithmen - Anwendung im Service Robotics||Authors:||Fragkopoulos, Christos||Supervisor:||Gräser, Axel||1. Expert:||Gräser, Axel||2. Expert:||Michels, Kai||Abstract:||
The thesis presents new approaches for autonomous motion execution of a robotic arm. The calculation of the motion is called motion planning and requires the computation of robot arm's path. The text covers the calculation of the path and several algorithms have been therefore implemented and tested in several real scenarios. The work focuses on sampling based planners, which means that the path is created by connecting explicitly random generated points in the free space. The algorithms can be divided into three categories: those that are working in configuration space(C-Space)(C- Space is the set of all possible joint angles of a robotic arm) , the mixed approaches using both Cartesian and C-Space and those that are using only the Cartesian space. Although Cartesian space seems more appropriate, due to dimensionality, this work illustrates that the C-Space planners can achieve comparable or better results. Initially an enhanced approach for efficient collision detection in C-Space, used by the planners, is presented. Afterwards the N dimensional cuboid region, notated as Rq, is defined. The Rq configures the C-Space so that the sampling is done close to a selected, called center, cell. The approach is enhanced by the decomposition of the Cartesian space into cells. A cell is selected appropriately if: (a) is closer to the target position and (b) lies inside the constraints. Inverse kinematics(IK) are applied to calculate a centre configuration used later by the Rq. The CellBiRRT is proposed and combines all the features. Continuously mixed approaches that do not require goal configuration or an analytic solution of IK are presented. Rq regions as well as Cells are also integrated in these approaches. A Cartesian sampling based planner using quaternions for linear interpolation is also proposed and tested. The common feature of the so far algorithms is the feasibility which is normally against the optimality. Therefore an additional part of this work deals with the optimality of the path. An enhanced approach of CellBiRRT, called CellBiRRT*, is developed and promises to compute shorter paths in a reasonable time. An on-line method using both CellBiRRT and CellBiRRT* is proposed where the path of the robot arm is improved and recalculated even if sudden changes in the environment are detected. Benchmarking with the state of the art algorithms show the good performance of the proposed approaches. The good performance makes the algorithms suitable for real time applications. In this work several applications are described: Manipulative skills, an approach for an semi-autonomous control of the robot arm and a motion planning library. The motion planning library provides the necessary interface for easy use and further development of the motion planning algorithms. It can be used as the part connecting the manipulative skill designing and the motion of a robotic arm.
|Keywords:||Service Robotics,Path planning,Robot arm, Configuration space||Issue Date:||11-Mar-2014||URN:||urn:nbn:de:gbv:46-00103710-19||Institution:||Universität Bremen||Faculty:||FB1 Physik/Elektrotechnik|
|Appears in Collections:||Dissertationen|
checked on Oct 26, 2020
checked on Oct 26, 2020
Items in Media are protected by copyright, with all rights reserved, unless otherwise indicated.