From 90392e35b4079ee7ccd657103df68e185e0aa60c Mon Sep 17 00:00:00 2001 From: The Open Journals editorial robot <89919391+editorialbot@users.noreply.github.com> Date: Wed, 16 Mar 2022 19:03:48 +0000 Subject: [PATCH] Creating 10.21105.joss.03967.jats --- joss.03967/10.21105.joss.03967.jats | 386 ++++++++++++++++++++++++++++ 1 file changed, 386 insertions(+) create mode 100644 joss.03967/10.21105.joss.03967.jats diff --git a/joss.03967/10.21105.joss.03967.jats b/joss.03967/10.21105.joss.03967.jats new file mode 100644 index 0000000000..2410e34674 --- /dev/null +++ b/joss.03967/10.21105.joss.03967.jats @@ -0,0 +1,386 @@ + + +
+ + + + +Journal of Open Source Software +JOSS + +2475-9066 + +Open Journals + + + +3967 +10.21105/joss.03967 + +TriP: A Python package for the kinematic modeling of +serial-parallel hybrid robots + + + +0000-0002-7825-3476 + +Baumgärtner +Jan + + + + + +Miller +Torben + + + + + +Heidelberg University + + + + +Independent Researcher + + + + +28 +10 +2021 + +7 +71 +3967 + +Authors of papers retain copyright and release the +work under a Creative Commons Attribution 4.0 International License (CC +BY 4.0) +2022 +The article authors + +Authors of papers retain copyright and release the work under +a Creative Commons Attribution 4.0 International License (CC BY +4.0) + + + +Python +robotics +inverse kinematics +kinematics +parallel manipulator +serial manipulator + + + + + + Summary +

​Robots can be classified according to their mechanical structure. + Serial mechanisms like robotic arms are mechanisms where each moving + part (called a link) is connected to only the one before and the one + after it. They are often used when a large workspace is required, + meaning the robot needs a long reach. In parallel mechanisms, the + links of the robot form loops causing them to be structurally stronger + and stiffer.

+

If both a large workspace and structural strength are required, + hybrids that contain both serial and parallel mechanisms are used. + While hybrid mechanisms combine the mechanical advantages of both + parallel and serial mechanisms, they also combine their modeling + disadvantages:

+ + +

Finding an explicit solution for either forward or inverse + kinematics is often impossible. Using numerical approaches instead + leads to complicated constrained optimization problems for both + forward and inverse kinematics.

+ + +

While serial mechanisms are very well supported by current + robotic frameworks, parallel mechanisms and hybrid mechanisms + especially are often not supported at all. A great overview of the + supported robot types for different robotic frameworks was + compiled by Kumar + (2019).

+ + +

TriP is a python package designed to close this gap using a modular + modeling framework akin to the one described by Kumar et al. + (2020). It + allows the modeling of arbitrary hybrid mechanisms and is capable of + calculating forward and inverse kinematics.

+

The calculations are performed using a symbolic framework. This + makes it easy for users to implement custom case-dependent + mathematical solvers.

+ + + Statement of Need +

While a huge number of researchers, such as Pisla et al. + (2013), Kanaan + et al. (2009), + and Zoss et al. + (2006), use + hybrid serial parallel systems, most modern kinematics frameworks + still lack support for such systems. Examples include OpenRAVE + (Diankov + & Kuffner, 2008), used in the MoveIt! stack + (Chitta + et al., 2012), the MATLAB robotics toolbox + (Corke, + 2017), Klampt + (Hauser, + 2021), and the inverse kinematics Python library + (Manceron, + 2021). While all these frameworks and libraries offer fast + computation of forward and inverse kinematics, they only support + branching serial mechanisms. This lack of support often leaves + developers with essentially two choices:

+

Either shoehorn their hybrid robots into a framework not designed + to handle them or be left to implement their own kinematic + solvers.

+

TriP is a lightweight and easy-to-use package for directly modeling + hybrid mechanisms and calculating their kinematics. Although TriP is + fast for a Python package, it is not built for robust hard-real time + control applications. Instead, it is aimed at researchers and + engineers who quickly want to build kinematic models to test their + mechanical designs. For this reason it also exposes a interface to the + nonlinear optimization and algorithmic differentiation tool CasADi + (Andersson + et al., 2019). This allows researchers to investigate the + kinematic properties of their robots, for example, by using CasADi to + find singular configurations or to compute manipulability ellipsoides. + Similar to MoveIt!, TriP also enables researchers to build their own + inverse kinematic solvers.

+ + + Overview +

TriP models robots using its Robot class. A + Robot object is made up of + Transformation and KinematicGroup + objects. The KinematicGroup objects are used to model + parallel mechanisms while the Transformation objects + model serial mechanisms. See Figure + 1 for + reference, where the links of each robot are colored according to the + KinematicGroup or Transformation it + belongs to, while joints are either light green or orange, depending + on whether they are actuated or not.

+ +

Different Hybrid Robot types and their object structure +

+ + +

Both KinematicGroups and + Transformations can be connected to form branching + mechanisms as indicated in Figure + 1.

+

Transformation objects implement homogeneous + coordinate transformations between coordinate frames. These can be + either dynamic or static with dynamic transformations implementing + joints. A few example joints can be seen in Figure + 2.

+ +

Sample Joints using the Transformation class +

+ + +

While all example joints use Euler angles in roll pitch yaw + convention to describe rotation, quaternions are also supported.

+

KinematicGroups model parallel mechanisms using + the abstraction approach described by Kumar et al. + (2020). This + approach models a parallel manipulator as a virtual serial manipulator + and a mapping that maps the virtual joint state to the true actuated + joint state of the parallel manipulator.

+

An illustrative example of this model is an excavator with two + hydraulic cylinders. Each cylinder is part of a parallel mechanism + resulting in two KinematicGroups. Both can be seen in + Figure 3, where one is + green and the other one is blue.

+ +

Excavator Arm build from two Groups (green and blue) +

+ + +

The abstraction approach models the excavator as a serial + manipulator where the joints are directly actuated. Using two mappings + to convert the state of the hydraulic cylinders to the state of the + joints and vice versa, it is possible to calculate both forward and + inverse kinematics. In this example, the mapping between cylinders and + joints can be expressed using trigonometry. Since an explicit + formulation of the mappings might not always be possible, TriP can + also compute the mapping by solving the closure equation of the + parallel manipulator.

+

TriP can generate symbolic representations of robots using CasADi + (Andersson + et al., 2019). This symbolic representation can be used to set + up a solver object that then solves the inverse kinematics. While the + library already implements a simple inverse kinematics solver, the + symbolic representation makes it easy to implement custom solvers.

+

All features of TriP are thoroughly documented with tutorials and + examples to help people get started.

+ + + + + + + + + HauserKris + + Kris’ locomotion and manipulation planning toolbox - Klamp’t + 2021 + 20210331 + https://github.com/krishauser/Klampt + + + + + + ManceronPierre + + IKPy - an inverse kinematics library aiming performance and modularity + 2021 + 20210710 + https://github.com/Phylliade/ikpy + + + + + + KumarShivesh + WoehrleHendrik + FernándezJosé + MuellerAndreas + KirchnerFrank + + A survey on modularity and distributivity in series-parallel hybrid robots + Mechatronics + 202004 + 68 + 10.1016/j.mechatronics.2020.102367 + + + + + + AnderssonJoel A E + GillisJoris + HornGreg + RawlingsJames B + DiehlMoritz + + CasADi – A software framework for nonlinear optimization and optimal control + Mathematical Programming Computation + Springer + 2019 + 11 + 1 + 10.1007/s12532-018-0139-4 + + + + + + ZossA. B. + KazerooniH. + ChuA. + + Biomechanical design of the Berkeley lower extremity exoskeleton (BLEEX) + IEEE/ASME Transactions on Mechatronics + 200605 + 11 + 10.1109/TMECH.2006.871087 + + + + + + KanaanDaniel + WengerPhilippe + ChablatDamien + + Kinematic analysis of a serial–parallel machine tool: The VERNE machine + Mechanism and Machine Theory + 2009 + 44 + 2 + 0094-114X + https://www.sciencedirect.com/science/article/pii/S0094114X08000517 + 10.1016/j.mechmachtheory.2008.03.002 + + + + + + PislaDoina + SzilaghyiAndras + VaidaCalin + PliteaNicolae + + Kinematics and workspace modeling of a new hybrid robot used in minimally invasive surgery + Robotics and Computer-Integrated Manufacturing + 2013 + 29 + 2 + 0736-5845 + https://www.sciencedirect.com/science/article/pii/S0736584512001196 + 10.1016/j.rcim.2012.09.016 + + + + + + DiankovRosen + KuffnerJames J. + + OpenRAVE: A planning architecture for autonomous robotics + 2008 + + + + + + CorkePeter I. + + Robotics, vision & control: Fundamental algorithms in MATLAB + Springer + 2017 + + + + + + KumarShivesh + + Modular and analytical methods for solving kinematics and dynamics of series-parallel hybrid robots + Universität Bremen + 2019 + + + + + + ChittaSachin + SucanIoan + CousinsSteve + + Moveit! [ROS topics] + IEEE Robotics & Automation Magazine + IEEE + 2012 + 19 + 1 + 10.1109/MRA.2011.2181749 + + + + + +