introduction to ros

University of Hamburg

MIN FacultyDepartment of Informatics

ROS Introduction

Introduction to ROS

Lasse Einig, Dennis Krupke, Florens Wasserfall

University of HamburgFaculty of Mathematics, Informatics and Natural SciencesDepartment of InformaticsTechnical Aspects of Multimodal Systems

April 7, 2016

L. Einig, D. Krupke, F. Wasserfall 1



Foundation ROS Introduction

Motivation

I Heterogeneity vs. HomogeneityI sensor types, actuators, ...I sensor model, kinematic chain, ...

I AbstractionI Algorithm re-usability

I 2D laser data mappingI object recognition

I DebuggingI simulation, data visualization, ...





Idea

I Robot Operating SystemI Meta operating systemI Open sourceI Hardware abstraction

I portabilityI simplification of sensors and actuators

I Recurring tasks already solvedI Navigation, data filtering, object recognition ...





Current State

I Multiple versions actively usedI may not be compatible to each otherI may not provide same libraries

I Linux (Ubuntu!)I Supports C/C++, Python, Java, Lisp, Octave ...

I Python for high level code/fast implementationI C/C++ for algorithms/computation

I Functions and algorithms already availableI May be difficult to findI Better than reimplementing




Structure ROS Introduction

ROS System

I ROS nodesI sensorsI actuatorsI logic

I ROS coreI Communication





ROS Node

I Discrete part of the systemI Specialized software/algorithmI Many ROS nodes per systemI Example:

I node gets imageI runs edge detection algorithm on itI provides found edges





ROS Core

I Central unit, also called ROS masterI nodesI sensorsI communication

I Coordination of nodesI Communication ManagementI Exactly one per systemI Transparent to the user





Communication

I MessagesI standardized data types

I TopicsI n:n communication

I Services and ActionsI 1:1 communication





Sensors

I ExplorationI LocalizationI DetectionI One node per sensor

I provide data as topicI abstract from hardware





System structure

applications fetch beercapabilitiesuniverse navigation; arm; grasping

libraries tf; pcl; opencv; bullet; eigentools rxgraph; rostopic; roslaunch

middlewaremain rosmaster; roscpp; rosbuild

I universe → robot centric, developed by communityI main → general tools, maintained by Willow Garage




Communication - Messages ROS Introduction

Messages

I Fundamental communication conceptI Description of data setI Data types

I ROSI general

I HeaderI time stampI identifier

$ rosmsg show -r robot msgs/Quaternion# xyz - vector rotation axis, w - scalar term (cos(ang/2))float64 xfloat64 yfloat64 zfloat64 w




Communication - Topics ROS Introduction

Topics

I Published by nodesI Unique identifierI AnonymityI Open subscriptionI Sensor data





Communication - Example

ROS master

camera viewer






ROS master

camera viewer

advertise(’images’)






ROS mastertopic:images

camera viewer







camera viewer

subscribe(’images’)







camera viewerimages(tcp)







camera viewer

publish(img)

images(tcp)







camera viewer

processorpublish(img)

images(tcp)







camera viewer


subscribe(’images’)

images(tcp)







camera viewer


images(tcp)

images(tcp)




Communication - Services ROS Introduction

Services

I 2 message typesI request and response

I Synchronous protocolI client sends requestI client waits for serverI server replies

$ rosservice type add two ints | rossrv showint64 aint64 b- - -int64 sum







camera viewer

processor

client

publish(img)

images(tcp)

images(tcp)







camera viewer

processor

client

publish(img)

wait for service(’process’)

images(tcp)

images(tcp)







camera viewer

processor

client

publish(img)

wait for service(’process’)Service(’process’,srv.Proc)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

ServiceProxy(’process’,srv.Proc)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

request

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

response

images(tcp)

images(tcp)




Communication - Actions ROS Introduction

ActionsI 3 message types

I goal and resultI optional feedback

I Asynchronous protocolI client sends goalI server may respond with feedbackI server delivers result

I Interruptible

# Define the goaluint32 dishwasher id # Specify which dishwasher we want to use- - -# Define the resultuint32 total dishes cleaned- - -# Define a feedback messagefloat32 percent complete







service:process

camera viewer

processor

client

publish(img)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

ActionClient

(’process’,ProcAction)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

wait for server(’process’)

images(tcp)

images(tcp)







service:process

camera viewer

processor

client

publish(img)

wait for server(’process’) ActionServer(’process’,ProcAction)

images(tcp)

images(tcp)







service:processaction:process

camera viewer

processor

client

publish(img)

images(tcp)

images(tcp)








camera viewer

processor

client

publish(img)

goal

images(tcp)

images(tcp)








camera viewer

processor

client

publish(img)

feedback(20%)

images(tcp)

images(tcp)








camera viewer

processor

client

publish(img)

feedback(40%)

abort

images(tcp)

images(tcp)








camera viewer

processor

client

publish(img)

feedback(60%)

images(tcp)

images(tcp)








camera viewer

processor

client

publish(img)

feedback(80%)

wait for result

images(tcp)

images(tcp)








camera viewer

processor

client

publish(img)

result

images(tcp)

images(tcp)




Simulations ROS Introduction

Simulations

I Important development toolI protects expensive hardwareI develop and test without robotI high-level test

I Simulates sensor dataI clean data

I TurtlesimI ROS learning tool

I GazeboI ROS simulator

I RVizI ROS data visualization





Turtle Sim

I Learning platformI 2D turtle

I moveI turnI draw

I CommunicationI ROS structure

Source: http://wiki.ros.org/turtlesim





Gazebo

I 3D rigid body simulatorI Simulates robots, environment and sensor data

Source: Lasse EinigL. Einig, D. Krupke, F. Wasserfall 20




RViz

I 3D visualization environmentI Different data can be shown

I Laser scan data, map, ...

Source: http://wiki.ros.org/turtlebot gazebo




Kinect ROS Introduction

Microsoft Kinect

I Motion sensing device for the XBox 360 by MicrosoftI Range camera technology by PrimeSenseI 3D depth information from infrared structured light

Source: https://msdn.microsoft.com





Kinect - technical details

I ResolutionI 640 x 480 @ 30 Hz colorI 320 x 240 @ 30 Hz depth

I FOV of 57° horizontally and 43° verticallyI Range ∼0.7 - 6 m

I up to 3.5 m realisticI Physical tilt range ±31°I Microphone array with 16 bit @ 16 kHz

I supports single speaker voice recognitionI OpenNI and Freenect drivers





Kinect - example

Source: TAMS, Uni Hamburg


introduction to ros

Documents