13 \ 4 Gi50 -.\-- M Ct-L- D~y\..c~A J...KD b0I f (2...0 4 Lf}-1Vl JVt ; AI'V

f y'U) C/ tVH1{ fV} / AJ ~ (["\ Y .i)

_L ~~L~u A-L \ \ A uTrw4§ ---"'-;-J;

Wte...--l~f)UJ(/ (21H11

~

f~5

~/V/f2A~

.Ypeu)CJvL,~rJ~ J/t-f {"SS

Ci -e n/erJLt C-L/tNy uA-9~5

-r-YLOtbDT)LE't \(.,:s1\...Jl[,uNS

CTit '1ft I ~-;-t) €..JC)

•.[~v:E fbc..L DF ~s»c())

~} ~ L--D seLecT~ M rvA-/'vP~ -

~.c.LJ ~fl tJ v'l-- It-B (3

\ \1\ 1"e-Yl- r- /7 c.-~

$eftf+\j/ou'tL -~A-~~.p_ rtD(,4.j S ~LO -C-+-t-S-

e.t> Ib Dl /Lt-rt-LTLt.e. . i=i> LLAuJ A- 1-\~ 0 /L uJ ~ LL)

Review of Robot Software Klafter pgs 523 -557

Last week we reviewed the ACL software for the Eshed Robot. This "Advanced Control Language" is

similar to Val as described by Klafter page 539-546. It is a programming language with extensions forrobotics.

The basic robotic programs allow:

Execution of Robot commands

Input and Output of data

Creation of programs

Simple file management.

ACL is also a multitasking system and can respond to inputs (Trigger and Wait).

We expect the software to do at least the following:

Overall control of the robotic system and Operator Interface

Robot control- Position control (coordinate transformations) with various motion types, speed

control, end effector control

Normal computer operations - programming, User I/O, program control (RUN, etc.), debugging

Sensor I/O (Analog) and Digital I/O and recognition of interrupts

Safety features - protect motors from overheating and mechanical parts from being driven

beyond their limits

More advanced:

Acceleration, force and torque control

Vision system with coordination with robot coordinates

Digital signal processing to filter inputs and reduce electrical noise from sensors

Obstacle avoidance (mobile robots or to protect robot)

Safety inputs from guards such as light or acoustic fences to protect personnel.

TECHNOLOGY NEWS

Robot as:A New Day forRobot DesignLee Garber

The Robot Operating System promises to make designing roboticsoftware easier and less expensive.

Robotics has undergonemany advances in recentyears, with robots gainingnew skills and the ability toundertake new tasks. For example,they are helping with surgery,and are even flying and executingcomplex instructions as drones.

However,robot use hasnot become as widespread asmany people have predicted.Even the development of indus-trial robots-formerly a brightspot-has stagnated.

One reason is that robots havenot had a standard platform onwhich developers could build.Instead, they typically have hadto start from scratch with eachproject. This has frequently madethe development process timeconsuming and expensive.

In addition, the lack of a stan-dard platform has frequently keptdifferent types of robotic softwarefrom being able to communicatewith one another, which has limiteddevelopers' ability to use multipleapplications within a single robot.This has also kept engineers frombeing able to reuse robotics soft-ware and otherwise collaborate.

This may all be about to change,though, as an increasing number

16 COMPUTER

of companies and researchers areworking with the open source RobotOperating System (ROS)platform-first released in 20tO-to provide....-software for their robots. And asthis has occurred, the pace of robot-ics innovation has increased.

Despite its name, ROSisn't justan operating system, althoughit includes some OS functional-ity. It is also a framework thatprovides software modules forperforming typical robot activi-ties such as object recognition.The technology also helps controland coordinate the modules.

"ROSis a software frameworkthat simplifies the task of writingcomplex robot software. With ROS,it is possible to quickly grab state-of-the-art software for many aspectsofthe system, which frees up R&Dtime for the novel aspects of aparticular project," said MorganQuigley,who developed the firstversion of ROSand is now chiefarchitect ofthe Open SourceRobotics Foundation (OSRF).

Proponents hope ROSwill enabledevelopers to flexibly and inexpen-sively create more capable robots.

However, the technology faceschallenges before it can becomethe dominant robotics platform.

Published by the IEEEComputer Society

A BRIEF HISTORYWhen QUigleyarrived at Stan-

ford University to study machinelearning for his doctoral program,he joined associate professorAndrew Ng's Stanford ArtificialIntelligence Laboratory (SAIL).Students there were working onthe Stanford Artificial Intelli-gence Robot.

STAIRreflected Ng's desire todesign a general-purpose robotthat could perform differenttypes of tasks, rather than theone narrow job that office andindustrial robots typically do.

To accomplish this, QUigleyreal-ized that SAILrequired a softwareframework that could integrateprograms from various studentsin a robot that would keep work-ing even if one of the applicationsfailed. In 2006, he designed a dis-tributed, peer-to-peer system calledSwitchyard to meet this need.

The next year, in a collabora-tion that led to ROS's creation,Quigley began working withw~ge, a research lab thatdeveloped personal robotics hard-ware and open source software.

Willow Garage, which has sinceclosed, released ROSversion 1.0,called BoxTurtle, in 20tO. There

0018-9162/13/$31.00 © 2013 IEEE