Nof S.Y. Springer Handbook of Automation

Подождите немного. Документ загружается.

Industrial Intelligent Robots 21.3 Intelligent Robots 355

this repetitive measurement using the 3-D vision

sensor.

Error Recovery

When building an automation system, it is desirable

to prevent errors that might stop the system before-

hand, but complete provisions to prevent errors are

impossible, and the occurrence of unexpected errors is

inevitable. For this reason, intelligent robots incorpo-

rate exception-handling functions that work in the event

of an error occurring. Exception handling analyzes the

cause of the error to recover the corresponding error

state. This prevents the fall in the operation rate of the

bin-picking system.

For instance, when a workpiece is taken out from

among those in the basket, the interference of the robot

and peripheral equipment might happen occasionally.

Then the robot controller with its high sensitivity col-

lision detection function, the details of which are shown

in Fig.21.13, detects the load suddenly added to the

robot by the interference, and stops the movement of

the robot instantaneously. The position and the posture

of the robot are retrieved, and it becomes possible for

the robot to continue to move afterwards. This function

prevents robot and peripheral equipment from being

damaged beforehand, and is useful for improving the

operation rate of the system.

The other 3-D vision method, the stereovision

method, uses images taken by two cameras, by match-

ing corresponding points on the left and right images

to calculate the objects’ 3-D position and posture

(Fig.21.14). Although it takes some time to match cor-

responding points, this method may be advantageous in

such a case where a mobile robot has to recognize its

surroundings, as there is no need to irradiate auxiliary

light as in the structured light method.

21.3.4 Force Sensors

A six-axis force sensor is used to give robots dexterity

by mounting it on the wrist of the robot arm. A force

sensor generally detects x-, y-, and z-direction forces

and each axis’ moment (a total of six degrees of free-

dom). The force sensor usually has a strain gauge on

the distortion part, which becomes distorted when ap-

plied with force or moment and makes it possible to

determine its value. Figure 21.15 shows an example

of the force sensor. The use of force sensors has en-

abled the robot to do such tasks as shaft ﬁtting and

gear phase matching for precision machine parts assem-

bly, as well as deburring and polishing, which require

Motor (6 axis)

Input torque

Torque in

collision

Torque in

motion

Yes

Dynamics

Collision

Threshold?

Motor rotational

speed (6 axis)

Robot dynamic model

Spring factor

Mass

Damping

factor

Fig. 21.13 Collision detection

Left camera imageRight camera image

Epipolar

line

Object

Left

Right

CCD camera

Correspondence

point matching

Fig. 21.14 Stereovision method

a certain pressure. Figure 21.16 shows the change of

the force and the moment in a peg-in-hole operation.

Fig. 21.15 Force sensor

Part C 21.3

356 Part C Automation Design: Theory, Elements, and Methods

Force

Moment

Peg

Hole

Fig. 21.16 Peg-in-hole with force sensor

Figure 21.17 shows an example of the assembly of the

crank mechanical unit of injection molding machine

with a force sensor. Figure 21.18 shows an example

of the assembly of small electronic devices with force

sensors.

21.3.5 Control Functions

Flexible Control Function of a Robot Arm

Conventional robots have difﬁculties in loading a cast-

ing workpiece accurately onto the chuck of a lathe or

the ﬁxture of a machining center due to deviance in

position or posture of the casting deriving from the

casting’s dimensional dispersion. In die-casting, there

Small electronic device

Force sensor

Fig. 21.18 Assembly of small electronic device with a force sensor

Force sensor

Crank

mechanical unit

Fig. 21.17 Assembly of a crank mechanical unit with

a force sensor

is also the risk that the conventional robot arm with

a workpiece does not comply with the ejector motion

when the workpiece is drawn out from the die-mold

by the ejector. The robot arm’s ﬂexible control func-

tion enables the robot to softly control its arm in the

Part C 21.3

Industrial Intelligent Robots 21.3 Intelligent Robots 357

Machining fixture

Workpiece

Fig. 21.19 Robot arm and workpiece comply with the ma-

chining ﬁxture so as to ﬁt the workpiece to the ﬁxture

designated direction in orthogonal coordinate systems.

It can thus accurately load a workpiece to the ma-

chine tool and draw out a workpiece from the die-mold

without any disturbing force in the die-casting opera-

tion. Figure 21.19 shows the ﬂexible control function

of robot arm. It does not use any force sensors but

rather the current of the servomotors that drive the robot

arm.

Coordinated Control Function

of Multiple Robots

Coordinated operation by multiple robots has been

made possible by synchronizing robots connected to

each other via the Ethernet. This function, for exam-

ple, enables multiple robots to carry a heavy workpiece

such as a car body in coordination, where the weight

of the workpiece exceeds the payload capacity of

one robot. Also, a ﬂexible system can be conﬁgured

as shown in Fig. 21.20, in which two robots rotate

a workpiece while gripping it, where the two robots

are used as rotating ﬁxtures, and another robot arc-

welds the workpiece in coordination with other robots’

motion.

Robot #2

Workpiece

Robot #1

Robot #3

Fig. 21.20 Coordinated arc welding system with three robots

Collision Detection Function

Humans may have a sense of afﬁnity with robots. How-

ever, robots are a highly rigid machine by nature, and

may cause heavy damage to human beings or other ma-

chines at the time of collision. In the latest robot control

technology, the drive power supply is immediately cut

when a robot collides with an object because it de-

tects the change in the servomotor current that drives

the robot arm. This minimizes the damage if a robot

collides with machines. Human beings must be sepa-

rated from the robot in space and/or time by such safety

means as fences.

21.3.6 Ofﬂine Programming System

Ofﬂine programming systems have greatly contributed

to decreasing the robot’s programming hours as PC per-

formance has improved. There is usually a library of

robot models in the ofﬂine programming system. The

data of workpiece shapes, which hasrecently been com-

piled by using 3-D CAD systems, is read into the ofﬂine

programming system. Peripheral equipment is often de-

ﬁned by using an easy shape generating function of the

ofﬂineprogramming system,or byusing 3-DCAD data.

As shown in Fig.21.21, after loading the workpiece’s

3-D CAD data into the ofﬂine programming system,

the robot motion program can easily be generated au-

tomatically, just by designating the deburring path and

the posture of the deburring tool on the PC display.

However, the robot motion program automatically

generated on the PC cannot immediately operate the

robot on the production ﬂoor, as the positional relation

of the robot and the workpiece on the PC display is

Part C 21.3

358 Part C Automation Design: Theory, Elements, and Methods

Office

CAD

data

LayoutAutomatic detection

of deburring lines

Setting of tool

orientation

Automatic conversion

to teaching points

Setting of approach

points

Generation of robot

motion program

Transfer

to robot

Fig. 21.21 Deburring program generation by ofﬂine programming

Factory floor

3-D vision

sensor

Fig. 21.22 Measurement of workpiece position and posture

by the vision sensor

slightly different from that of the real robot and work-

piece on the production site. The latest technology in

robot vision sensors calculates the real workpiece’s po-

sition and posture on the production ﬂoor, enabling an

automatically generated program to be corrected au-

tomatically, as shown in Fig.21.22, thus signiﬁcantly

decreasing the programming time for deburring.

21.3.7 Real-Time Supervisory

and Control System

The real-time supervisory and control system is a soft-

ware package to monitor and control the production

system’s real time, which is generally called SCADA

(supervisory control and data acquisition). As shown

in Fig. 21.23, the system enables an operator to gener-

ate screens interactively, by making use of modularized

functions of data collection, graphical drawing and data

analysis software, as well as standard communication

network and common database.Various productiondata

including machining data and cycle time can be con-

trolled and analyzed from an ofﬁce via the Ethernet,

which links the ofﬁce computers with the CNC machine

tools and robots on the production ﬂoor. The plant man-

ager can directly access the production outcome and

operational status of machine tools and robots in the

plant he/she is responsible for via internet or by using

a mobile phone even during his/her travels abroad, to

give precise and timely directions. For a related discus-

sion see Chap.23 on programming real-time systems

for automation.

There is also a system available with a function

to strongly support ﬁnding causes of temporal system

stoppages during the system operation on the produc-

tion ﬂoor.

Part C 21.3

Industrial Intelligent Robots 21.4 Application of Intelligent Robots 359

Robot status RoboDrill status

Fanuc cimplicity/roboguide

Fig. 21.23 Real-time supervisory and control system

21.4 Application of Intelligent Robots

21.4.1 High-Speed Handling Robot

In the food and pharmaceutical handling ﬁeld, there

has been a delay in robotic automation compared with

heavy workpiece handling, as the goods handled in this

ﬁeld are relatively lighter, and as there is a stronger

requirement for higher-speed and continuous handling

operation. Figure 21.24 shows an example of the high-

speed handling by an intelligent robot, which can

operate continuously at high speeds, and is clean, wash-

able, and chemical proof. This robot adopts the dual

drive and torque tandem control method as shown in

Fig.21.25, the ﬁrst case in robots, to achieve high-speed

and continuous operation. Each basic axis has two mo-

tors, and by optimally controlling these motors, high

acceleration/deceleration and continuous operation for

high duty performance are achieved.

Using the visual tracking function, which combines

a vision sensor and a tracking function, can increase

handling efﬁciency. With the vision function built into

the robot controller, the vision system has become

highly reliable for use on the production ﬂoor. With

its vision sensor the robot recognizes the position of

a workpiece coming on the conveyer and compensates

the robot motion trajectory by comparing the work-

piece’s position with the conveyor speed data received

from the pulse coder. Also, the system allocates han-

dling operations among multiple robots in order to

increase handling efﬁciency. Figure 21.24 shows the

visual tracking function by three high-speed handling

robots, which can handle 300parts/min.

Vision sensor

Fig. 21.24 Visual tracking system by high-speed handling robots

Part C 21.4

360 Part C Automation Design: Theory, Elements, and Methods

Speed control

Position control

Current control

Torque command

Main motor

Sub motor

Fig. 21.25 Dual drive and torque tandem control

21.4.2 Machining Robot Cell –

Integration of Intelligent Robots

and Machine Tools

Generally, the machining system can reduce machining

costs by operating long hours continuously. The history

of machining system is summarized in Fig.21.26.

1. In the 1980s, a 24h continuous operation was

achieved by introducing a CNC machine tool sys-

tem equipped with pallet magazines (Fig.21.26a).

2. In the 1990s, a 72h continuous operation including

weekends was achieved by introducing a machining

system equipped with a large-scale multilayer pallet

stocker (Fig.21.26b).

3. In the 2000s, a robot cell system was developed

in which an intelligent robot directly loads work-

pieces to the machining ﬁxtures of the machining

center, achieving 720 h per month, or 24h for

30days, continuous operation. The intelligent robot

1980s

1990s

2000s

Pallet magazine Large-scale

multilayer pallet

stocker

Intelligent robot

– robot cell –

24 hours continuous

operation

72 hours continuous

operation

720 hours continuous

operation

1st Generation 2nd Generation 3rd Generation

a) b) c)

stocke

Fig. 21.26a–c A trend in systems for long hours of continuous ma-

chining

Intelligent robot

RoboDrill

Fig. 21.27 Mini robot cell. Decreases machining cost of

long hours of continuous machining drastically

compensates the deviation in its gripping of the

workpiece, resulting from the dimensional varia-

tion of the casting, with its 3-D vision sensor. Also,

the workpiece is loaded to the machining ﬁxture

with precision, because the workpiece is pressed

softly to the surface of the ﬁxture by controlling

the robot arm softly as shown in Fig. 21.19. The

robot cell can substantially reduce labor and ma-

chining costs, as well as initial capital investment

(Fig.21.26c).

Figure 21.27 shows a mini robot cell comprised of

a CNC drill and an intelligent robot, in which the intel-

ligent robot loads and unloads workpieces to and from

the CNC drill, and measures the dimensions of the ma-

Fig. 21.28 Top mount loader robot

Part C 21.4

Industrial Intelligent Robots 21.4 Application of Intelligent Robots 361

Fig. 21.29 Big robot

chined workpieces. An operator carries a basket ﬁlled

with workpieces in front of the robot. The intelligent

robot with a vision sensor picks up a workpiece from

among those randomly placed in the basket and loads

it to the drill. This robotic system eliminates dedicated

workpiece supply equipment, as well as the manual

work by human operators to array workpieces on it.

Mini robot cells are available from the minimum

conﬁguration of one machine tool and one intelligent

robot to a system with multiple machine tools and

robots, meeting the various requirements of users. In

the case where multiple machine tools work on a work-

piece for different machining processes, there is also

a system in which a mobile intelligent robot runs be-

tween the machine tools to transfer the workpiece as

showninFig.21.28.

Until now, a heavy workpiece of about 1000kg had

to be handled by the coordination of more than two

robots, and workers had to use the crane to handle the

heavy workpiece. A big intelligent robot, whose pay-

load is about 1000kg appeared recently. This robot can

simplify the robot cell as shown in Fig.21.29.

21.4.3 Assembly Robot Cell

Intelligent robots are also expected to take an active

role in the assembly job, which comprises as large

Mini robot Intelligent

robot

Intelligent robot

Force

sensor

Vision

sensor

Fig. 21.30 Assembly robot cell

a part of the machine industry as the machining job.

The intelligent robot can perform highly accurate as-

sembly jobs, picking up a workpiece from randomly

piled workpieces on a tray, assembling it with the

ﬁtting precision of 10μm or less clearance with its

force sensor. Figure 21.30 shows an assembly robot

cell in which intelligent robots are assembling mini

robots.

Part C 21.4

362 Part C Automation Design: Theory, Elements, and Methods

21.5 Guidelines for Installing Intelligent Robots

The following items are guidelines that should

be examined before an intelligent robot is intro-

duced.

21.5.1 Clariﬁcation of the Range

of Automation by Intelligent Robots

Though the intelligent robot loads workpieces to the

machine tool and assembles parts with high accuracy,

it cannot do everything a skilled worker can do. For in-

stance, the task such as the assembly of a ﬂexible thing

belongs to a ﬁeld in which humans are more skillful

than intelligent robots. It is necessary to clearly sep-

arate the range that can be automated from the range

that relies on skilled workers before the introduction of

intelligent robots.

21.5.2 Suppression

of Initial Capital Investment Expense

The effect of the introduction of intelligent robots arises

if the peripheral equipment can be simpliﬁed by mak-

ing use of ﬂexibility that is one of the major features

of intelligent robots. This effect weakens if the initial

capital investment expense is not suppressed due to an

easy increase of the expense for peripheral equipment

compared with that of the automation that does not use

intelligent robots.

21.6 Mobile Robots

Figure 21.31 shows an autonomous mobile cleaning

robot. The robot is mainly used for ﬂoor-cleaning in

skyscrapers. It moves between stories by operating the

elevator by itself, and cleans the ﬂoors at night. It au-

Main controller

Gyroscope

Optical transmitter

Filter

Obstacle sensor

Bumper

Drive wheel

Obstacle sensor

Blower motor

Motor controller

Cleaning

apparatus driver

Running indicator

light

Suction nozzle

with power brush

Fig. 21.31 Autonomous mobile cleaning robot (source: Fuji Heavy Industries Ltd.)

tonomously returns to the start position after it has

worked.More than 10 cleaning systems using the robots

have already been introduced into several skyscrapers in

Japan.

Part C 21.6

Industrial Intelligent Robots References 363

21.7 Conclusion

The intelligent robot appeared on the factory ﬂoor

at the beginning of 2000 and has vision and force

sensors. It can work unattended at night and during hol-

idays because it reduces manual preparations such as

arrangement of workpieces and/or the necessity of sys-

tem monitoring compared with the conventional robot.

Thus, the use efﬁciency of equipment rises, machin-

ing and assembly costs can be reduced, and the global

competitiveness of a product can be improved.

The industrial intelligent robots still have tasks

in which they cannot compete with skilled workers,

though they have a high level of skills, as has been

explained so far. The assembly task of ﬂexible objects

such as wire-harnesses is one such task. There are sev-

eral on-going research and development activities in the

world to solve these challenges. One idea is to automate

such a task completely, and another is to do it partially.

In the latter case, robots and skilled workers work to-

gether; robots assemble mechanical parts and skilled

workers assemble ﬂexible parts, for example. In any

case, the degree of cooperation between humans and

robots will increase in the near future.

21.8 Further Reading

•

Y. Bar-Cohen, C. Breazeal: Biologically Inspired In-

telligent Robots (SPIE Press, Bellingham 2003)

•

G.A. Bekey: Autonomous Robots: From Biologi-

cal Inspiration toImplementation and Control (MIT

Press, Cambridge 2005)

•

J.M. Holland: Designing Autonomous Mobile

Robots: Inside the Mind of an Intelligent Machine

(Newnes, Amsterdam 2003)

•

S.C. Mukhopadhyay, G.S. Gupta: Autonomous

Robots and Agents (Springer, New York 2007)

•

S.Y. Nof (Ed.): Handbook of Industrial Robotics

(Wiley, New York 1999)

•

R. Siegwart, I.R. Nourbakhsh: Introduction to Au-

tonomous Mobile Robots (MIT Press, Cambridge

2004)

•

P. Stone: Intelligent Autonomous Robotics: A Robot

Soccer Case Study (Morgan Claypool, San Rafael

2007)

References

21.1 D.E. Whitney: Historical perspective and state of

the art in robot force control, Proc. IEEE Int. Conf.

Robot. Autom. ICRA (1985)

21.2 T. Suehiro, K. Takase: Skill based manipulation

system, J. Robot. Soc. Jap. 8(5), 551–562 (1990)

21.3 M.T. Mason: Compliance and force control for com-

puter controlled manipulators, IEEE Trans. Syst.

Man Cybern. 11(6), 418–432 (1981)

21.4 N. Hogan: Impedance control, Part 1–3, Trans. ASME

J. Dyn. Syst. Meas. Control 107, 1–24 (1985)

21.5 J.J. Craig, P. Hsu, S. Sastry: Adaptive control of

mechanical manipulators, Int. J. Robot. Res. 6(2),

16–28 (1987)

21.6 B. Yao, M. Tomizuka: Adaptive coordinated control

of multiple manipulators handling a constrained

object, Proc. IEEE Conf. Robot. Autom. (1993)

21.7 S. Inaba: Assembly of robots by AI robot, Proc.

22nd IEEE IECON (Int. Conf. Ind. Electron. Control

Instrum.), Vol. 1 (1996) pp. xxxvii–xl

21.8 S. Sakakibara, A. Terada, K. Ban: An innovative

automatic assembly system where a two-armed

intelligent robot builds mini robots, Proc. 27th ISIR

(1996) pp. 937–942

21.9 S. Sakakibara: Intelligent assembly system suited

for module assembly, Proc. 30th ISR (1999) pp. 385–

390

21.10 S. Sakakibara: The role of intelligent robot in man-

ufacturing system of the 21st century, Proc. 32nd ISR

(2001)

21.11 K. Hariki, K. Yamanashi, K. Otsuka, M. Oda: In-

telligent robot cell, FANUC Tech. Rev. 16(1), 37–42

(2003)

Part C 21

“This page left intentionally blank.”