See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323709086

Mobile manipulation task simulation using ROS with MoveIt

Conference Paper · July 2017

DOI: 10.1109/RCAR.2017.8311930

CITATIONS READS

4 3,625

3 authors:

Hao Deng Jing Xiong

Chinese Academy of Sciences Chinese Academy of Sciences
11 PUBLICATIONS   60 CITATIONS    68 PUBLICATIONS   297 CITATIONS   

SEE PROFILE SEE PROFILE

Zeyang Xia
Chinese Academy of Sciences
78 PUBLICATIONS   398 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Motion planning and Control View project

Incremental training of regression and dimensionality reduction neural networks View project

All content following this page was uploaded by Hao Deng on 09 August 2018.

The user has requested enhancement of the downloaded file.

 Proceedings of The 2017 IEEE International
Conference on Real-time Computing and Robotics
July 14-18, 2017, Okinawa, Japan

Mobile Manipulation Task Simulation using ROS with MoveIt

Hao Deng1,2 , Jing Xiong1 and Zeyang Xia1,2,*

Abstract— Visualized simulation has always been one of the
most flexible techniques for quick verification of prototype or
algorithms. In this paper, we illustrate a simple and effective
method of how to implement the simulation of mobile manipu-
lation task using Robot Operation System (ROS) with MoveIt.
First, we propose a quick and modular architecture which
includes the entire configuration process of system modeling,
kinematics, control, and visualization. Then, we demonstrated
its application on simulation of a mobile manipulation task in
the automatic production line. Experimental results have shown
that the proposed architecture achieves an easy configuration, a
good integration and functional visualization for manipulation
simulation
Index Terms— ROS, MoveIt, mobile manipulation, simula-
tion.
hardware diversity. Early examples include simulations built
with Adams and Matlab for handling of robot movement
and dynamic analysis [6, 7]. Later work also used OpenGL
to develop their own simulator, or even with Open Dynamic
Engine (ODE) integrated for dynamics [8, 9]. Previous work
and development has been presented in details can be found
in [10]. This early work has been followed by an abundance
of research considering the functional integrity. Each of these
frameworks was designed for particular purposes, or perhaps
in response to perceived weaknesses of other frameworks.

I. INTRODUCTION
Simulation techniques have played a critical important
role in robotic development for quick verification of new
prototype, algorithms or applications, and especially system
performance optimization [1]. A robotic simulator is capable
of emulating the motion of robots and all objects in a
virtual work envelope, but without physically depending on
an actual complete hardware system, thus saving cost and
time [2].
This has, during past few years, led to increased interests
for robotic simulation. In the basic form, modeling and rend-
ing of a robot and its environment has been well studied [3-
5]. Robot manufactures, such as ABB, KUKA, Staubli, et al.
have developed simulation suits for their own robots (Fig.1(a) Fig. 1. Examples of exsiting roboic manilulation simulation platforms. (a)
KUKA WorkVisual [14], (b) Microsoft Robotic Developer Studio (MRDS),
for KUKA WorkVisual), which allows for robotics programs (c) Open Robotics Automation Virtual Environment (OpenRAVE), (d) Rviz
to be conveniently written, debugged off-line and tested on in Robot Operation System (ROS).
an actual robot, but in more complex applications, third-party
hardware and services are difficult to be integrated, sensor- There also exists a variety of robot simulation platform
based actions or physical engines, for example. developed in open sourced and running as a community
Recently, there is an increasing trend of creating a wide project. Microsoft Robotic Developer Studio (MRDS) [11]
variety of frameworks to manage the system complexity and is a Windows-based environment for robot control and sim-
This work was supported by the National Science Foundation of China ulation. MRDS can handle a wide variety of robot hard-
(61403368), National Key Project of Research and Development Plan ware, include support for packages to add other services to
of China Inter-governmental Cooperative Science and Technology Project achieve complex behaviors. MRDS is likely to be a powerful
(2016YFE0128000), Guangdong Natural Science Foundation for Distin-
guished Young Scholars (2015A030306020), Major Project of Guangdong and widely used platform before it has not been updated
Province Science and Technology Department (2014B090919002), Shen- or patched since 2012. Open Robotics Automation Virtual
zhen High-level Oversea Talent Program Grant (KQJSCX20160301144248), Environment (OpenRAVE) [12] provides an environment
Youth Innovation Promotion Association, Chinese Academy of Sciences
(2015301) and Shenzhen Research Project (GJHS20160331185913023, for development of motion planning algorithms in robotics
GJHS20160331190459402). applications, which focuses on offering interfaces and im-
1 Shenzhen Institutes of Advanced Technology, Chinese Academy of
plementations of motion planning algorithms. OpenRAVE
Sciences, Shenzhen, China
2 CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems, is still actively being maintained and has been integrated
Chinese Academy of Sciences, Shenzhen, China in many simulators. Robot Operation System (ROS) [13]
∗ Author to whom correspondence should be addressed: Shenzhen Insti-
is a robotic middleware providing software frameworks for
tutes of Advanced Technology, 1068 Xueyuan Avenue, Shenzhen University
Town, Shenzhen, China. Email: zy.xia@siat.ac.cn; Phone: +86- the development of robotic software. ROS contains many
755-8639-2218 open source implementations of common robotics function-

978-1-5386-2034-2/17/$31.00 © 2017 IEEE 612

ality and algorithms. These open source implementations
are organized by packages included in ROS distributions,
while others may be developed by individuals and distributed
through code sharing site such as GitHub. And in recent
years, ROS has become the most widely used platform for
robotic research.
Even though there are a variety of optional platforms
for robotic simulation, developing software for manipulation
task is still difficult, particularly as the scale and complexity
of the hardware system continues to grow. Requirement
of system integration, code reuse, compatibility for new
algorithms or strategies makes even professional users still
take long time to configuration complex function modules.
This paper is aimed to present the idea that how to easily
understand the and make use of ROS to develop your
simulation software for a specific manipulation task. In our
work, we propose quick and modular architecture which
includes the entire configuration process of system modeling,
kinematics, control, and visualization. A case study of mobile
manipulation task simulation in automatic production line
demonstrates standard procedures of creating a simulation
of mobile manipulator and illustrating the methods of how
to implement robot molding, planning and control.
This paper is organized as follows. Section II provides an
overview of ROS concepts. Section III depicts procedures of
simulation system implementation. Section IV demonstrates
its application. Section V summarizes our work and starts a
discussion about simulation of robotic manipulation task.
II. RELATED WORK
We first briefly introduce the basic concepts of ROS.
Fig. 2. ROS graph concepts. (a) Core computation graph level, (b) messages
publish and subscribe communication between nodes to topics, (c) service
request and response communication between server and client, (d) action
communication between server and client.
ROS was initially developed in 2007 by Stanford Artificial
Intelligence Laboratory and its original philosophy is to
design a flexible and dynamic software system [13]. Thus
makes ROS a collection of software frameworks for robotic
development. Usually ROS works as a robotic middleware
613
between software applications and actual operation system,
Linux for instance, providing standard operation system ser-
vices such as hardware abstraction, low-level device control,
commonly functionality implementation, message-passing,
and package management.
The computation graph is the peer-to-peer network of
ROS processes that are processing data together. roscore,
according to the literal, is the core of the system providing
collection of nodes and programs[15]. Currently roscore
contains three main functional module suits, as shown in
Fig. 2 (a), which are ROS Master, Parameter Server and
rosout logging node. ROS Master provides naming and
registration to nodes in the ROS system. Each individual
node are given an unique name and assigned with a systemic
pid. ROS node can locate each other and build peer-to-peer
communications in three modes, as given in Fig 2 (b-d),
which are the message publish/subscribe mode (one way),
the service server/client mode (one way but with feedback),
and the action server/client mode (bidirectional).
Based on the above concepts, a general robotic simulation
architecture can be rough classified into four key functional
modules, i.e., modeling, kinematics, control, and visualiza-
tion. The relationship of modules are illustrated as Fig. 3.
Fig. 3. General robotic simulation architecture and module relationship.
III. METHOD
To develop a robotic applications, it is required to build
at least the above four packages, and for a long time, ROS
users need to original build or modify community shared
repository for the customized robotic system. In our work, we
employ the state of art software tool MoveIt [16] to realize
an easy-to-use procedures for developing advanced robotics
applications.
A. System Modeling and Description
Robotic system can be modeling in tree structures. In ROS,
an Unified Robot Description Format (URDF) is designed
to describe the system model configuration. URDF file is a
XML specification recorded the whole physical information
of link, joints, properties, actuators and sensors. In our study,
we used a community toolkit sw urdf exporter, which is a
SolidWorks add-in, to conveniently export of CAD models
into URDF file [17]. As shown in Fig. 4, the explore can
create parts of files in final description packages, including
a directory for meshes, texture and robots urdf file. And also
it can automatically determine the proper joint type, joint
transforms and axes.
Fig. 4. Automatic robot description URDF generation. (a) using the
sw urdf exporter add-in, (b) link tree for mobile manipulator.
This toolkit can save lots of developing time, but for
ideal simulation results, the exported package still need some
minor modification. Here we put material properties, deleted
useless dummy links and reduced the accuracy of collision
models for a better collision checking performance.
The transmission is an extension to the urdf file, describing
the relationship between an actuator and its joint, which
is the necessary configuration for robot controller. In our
precious work [18], we have illustrated how to implement
robot control. For laser sensor, Hokuyo, configuration, we
used the community shared urg node package. Thus we
can realize a complete file configuration describing a mobile
manipulation platform. And the models can be shown in Rviz
as given in Fig. 5.
Fig. 5. Mobile manipulator model and render in simulator Rviz.
B. Planning and State of Art Algorithms Integration
MoveIt is a ROS-based software for robotic manipulation,
providing easy-to-use platform to develop robotic applica-
tions[16]. In our study of mobile manipulation, we use ROS
with MoveIt framework to develop customized planner and
integrate state of are motion planning algorithms.
614
The motion planning plugin in MoveIt is based on an Open
Motion Planning Library (OMPL) [19], and the interface
to motion planners is through a ROS Action or service.
OMPL provides foundational libraries support so that dif-
ferent planners with multiple algorithms can be integrated,
making planner module easily extensible. In Fig. 6 we
have implemented three RRT variant planners developed for
mobile manipulation planning.
Fig. 6. User developed planner integrated in MoveIt.
The loaded planner could generate a desired trajectory in
response to the initial and goal configuration, and also obeys
the geometric and kinematic/dynamic constrains.
Motion planner works with kinematic solver and collision
checker. MoveIt allows us to write our own inverse kinematic
algorithms. The RobotState class in MoveIt can get forward
kinematics and find jacobians. And we implement the inverse
kinematics solver using the TracIK [20] for its efficient and
stable inverse kinematic performance. Collision checking in
MoveIt is mainly carried out using the FCL library [21],
and supports for different types of objects including meshes,
primitive shapes and octomap objects. Motion planner will
eventually generate collision-free path and in our study, the
result of a 7-DOFs Sawyer manipulator is given in Fig. 7.
Fig. 7. Motion planing for 7DOFs Sawyer manipulator.
C. Control and Hardware Interface Implementation
To drive virtual or physical robot to move, control sys-
tem need talk to robot controller through ROS topics and
actions. It communicates with the controllers to get current
robot states, instant images and other sensor data. And the
ROS control also execute high-level commanders and sends
signals to low-level actuators. Architecture of this module is
illustrated as Fig. 8.
 Hardware interface is the bridge between hardware devices
and ROS application. With hardware drives, commands can
be sent to controller through ROS messages, services or
actions. ROSIndustrial [22] provide hardware oriented func-
tionality to connect robots, cameras and sensors with ROS.
And provide supporting software for many vendor platforms.
Each platform requires different setup and configuration
steps in order to be used with ROS-Industrial. In our case of
study, Sawyer robot, Robotiq gripper, p3at mobile base and
sensors are well supported and easily implemented. In our
previous work [23], we have demonstrated how to develop
the general interfaces for customized hardware and protocols.
Fig. 8. Robot control module architecture.
D. Visualization and Task Simulation
For manipulation simulation, above modules are working
under the hood, only the simulator GUI or visualization
toolkit can be seen by users [24]. With all the models loaded,
planner and control started, we use Rviz to visualize the
simulation given in Fig. 9.
Fig. 9. Mobile manipulation simulation scene in Rviz.
IV. RESULTS AND DISCUSSION
In order to demonstrate procedures of creating simulation
and illustrate the proposed methods of how to implement, we
conduct a case study of mobile manipulation task simulation
in automatic production line.
The mobile manipulator need navigate in the unstructured
industrial environment with point cloud map building from
615
the laser sensor, then pick a workpiece and place at work-
holding position of the machining center.
Fig. 10. Snapshots of the mobile manipulation task simulation.
ROS has been becoming powerful tool for robotic research
or system development. It is super easy to integrate different
functional packages from multiple projects, or even borrow
from other simulation platform, such as Trac-IK, the ik plu-
gin from OpenRAVR. With MoveIt, many foundational work
can be done automatically. Clearly using ROS and its com-
munity resources, researchers can shorten the development
cycle time, and take efforts for new features implementation,
which will in turn enhance the extensible ability into new
applications. Since the intention of this paper is to guide
users from ROS concept to practical applications, and may
also hope to provide available solutions and resources in
simulation of the robotic manipulation task, technical details
have not gone deeper.
And there is also a clear trend that ROS development
will have moved from the early extensive functionality into
promising integrated solution.
R EFERENCES
[1] Torres-Torriti M, Arredondo T, Castillo-Pizarro P. Survey and compar-
ative study of free simulation software for mobile robots [J]. Robotica,
2016, 34(04): 791-822.
[2] Qian W, Xia Z, Xiong J, et al. Manipulation task simulation using
ros and gazebo [C]. Robotics and Biomimetics (ROBIO), 2014 IEEE
International Conference on. IEEE, 2014: 2594-2598
[3] Biber P, Andreasson H, Duckett T, et al. 3D modeling of indoor
environments by a mobile robot with a laser scanner and panoramic
camera [C]. Intelligent Robots and Systems, 2004.(IROS 2004). 2004
IEEE/RSJ International Conference on. IEEE, 2004, 4: 3430-3435..
[4] Pepper C, Balakirsky S, Scrapper C. Robot simulation physics val-
idation [C]. 2007 Workshop on Performance Metrics for Intelligent
Systems. ACM, 2007: 97-104.
[5] Karakaya S, Kucukyildiz G, Ocak H. A New Mobile Robot Toolbox
for Matlab [J]. Journal of Intelligent & Robotic Systems, 2017: 1-16.
[6] Yi Z, Wenshi S. Dynamics Analysis of 6-DOF Robot based on
ADAMS [C]. 2017 International Conference on Mechatronics Systems
and Control Engineering. ACM, 2017: 6-9.
[7] Krishnamurthy D A. Modeling and Simulation of Skid Steered Robot
Pioneer 3AT [J]. 2017.
[8] Laue T, Spiess K, Rfer T. SimRobota general physical robot simulator
and its application in robocup [C]. Robot Soccer World Cup. Springer
Berlin Heidelberg, 2005: 173-183.
 [9] Hugues L, Bredeche N. Simbad: an autonomous robot simulation
package for education and research [C]. International Conference on
Simulation of Adaptive Behavior. Springer Berlin Heidelberg, 2006:
831-842.
[10] Kramer J, Scheutz M. Development environments for autonomous
mobile robots: A survey [J]. Autonomous Robots, 2007, 22(2): 101-
132.
[11] jackson J. Microsoft robotics studio: A technical introduction [J]. IEEE
Robotics & Automation Magazine, 2007, 14(4).
[12] Diankov R, Kuffner J. Openrave: A planning architecture for au-
tonomous robotics [J]. Robotics Institute, Pittsburgh, PA, Tech. Rep.
CMU-RI-TR-08-34, 2008, 79.
[13] Quigley M, Conley K, Gerkey B, et al. ROS: an open-source Robot
Operating System [C]. ICRA workshop on open source software. 2009,
3(3.2): 5.
[14] https://www.kuka.com/en-us/products/
robotics-systems/software/system-software/kuka_
systemsoftware/kuka-work-visual.
[15] http://wiki.ros.org/ROS/Concepts.
[16] http://moveit.ros.org/.
[17] http://wiki.ros.org/sw_urdf_exporter.
[18] Qian W, Xia Z, Xiong J, et al. Manipulation task simulation using
ros and gazebo [C]. Robotics and Biomimetics (ROBIO), 2014 IEEE
International Conference on. IEEE, 2014: 2594-2598.
[19] http://ompl.kavrakilab.org/.
[20] P. Beeson and B. Ames. TRAC-IK: An Open-Source Library for
Improved Solving of Generic Inverse Kinematics. In International
Conference on Humanoids, 2015.
[21] J. Pan, S. Chitta, D. Manocha, FCL: A general purpose library for
collision and proximity queries [C]. In International Conference on
Robotics and Automation (ICRA), 2012.
[22] Edwards S, Lewis C. Ros-industrial: applying the robot operating sys-
tem (ros) to industrial applications [C]. IEEE International Conference
on Robotics and Automation, ECHORD Workshop. 2012.
[23] Deng H, Xia Z, Weng S, et al. A motion sensing-based framework for
robotic manipulation [J]. Robotics and Biomimetics, 2016, 3(1): 23.
[24] Zaman S, Slany W, Steinbauer G. ROS-based mapping, localization
and autonomous navigation using a Pioneer 3-DX robot and their
relevant issues [C]. Electronics, Communications and Photonics Con-
ference (SIECPC), 2011 Saudi International. IEEE, 2011: 1-5.

616

View publication stats