Robot simulation and off-line

programming
by S. F. Chan, Prof. R. H. Weston and K. Case
Loughborough University of Technology

Robot simulation and off-line programming seem likely to have an 0 It may be possible to create better
increasingly important role in the evolution of C I M facilities. This quality programs using more intelligent
article considers the capabilities and shortcomings of the current languages than those identified above.
0 The use of a teach pendant may
generation of off-line programming and graphical simulation systems. raise operator safety issues.
Methods of overcoming problems associated with calibration, lack of
‘user friendliness’ and integration within manufacturing systems Commercially available robot
are discussed through reference to research studies at simulators
Loughborough University.
A number of robot simulators have
become available in the world market-
place (Refs. 3-6), including GRASP,
Introduction relating the process and product infor- McDonnell Douglas Robotics Software,
mation to investment and resource plan- AutoSi mulations, RoboTeach, AUTO PASS
Industrial robot systems have conven- ning criteria. Without computer assis- and ROBCAD. These CADCAM packages
tionally been designed and programmed tance in this procedure it i s highly likely provide a set of modelling and simu-
as stand-alone flexible manufacturing that it will only be possible to find opti- lation tools which can be used to
mac h i nes. However, severe Ii m itat ions mal solutions for the simplest of manu- represent graphically a robot manipu-
in this approach are now being encoun- facturing tasks lator and its attendant equipment, and
tered as manufacturing companies seek Having designed and assembled a hence simulate a manufacturing task.
to become more competitive through robot system, it is necessary to create The use of such packages can allow the
evolving computer integrated manufac- robot task programs which determine manufacturing engineer to try several
turing (CIM) facilities. the axial movements that the robot solutions for robotic cells before
There are some parallels here with the manipulators will perform in relation to purchasingany equipment or to investi-
development of part programming for mechanisms in the workplace and in gate alternative uses of existing cells.
numerically controlled (NC) machines, response to sensory feedback. The use Hence these workplace design tools can
where initial requirements were for of a centralised CADCAM package to be used to improve the choice and lay-
methods of programming single, rela- create robot task programs (commonly out of robot systems, reduce set-up
tively simple machines through the referred to as a method of off-line pro- costs, reduce installation times and
facilities of their own controllers (on- gramming) is not a simple exercise. The improve system performance: certain
line programming). Advances in com- variability and complexity of the kine- manufacturers claim that ‘engineers can
puter-aided design and manufacturing matics of different robot manipulators design and lay out robot cells up to 70%
(CADCAM) and the advent of computer and workplaces raises new issues when faster through the use of a robot
and direct numerical control (CNC/DNC) compared with the off-line programming simulator’ (Ref. 4).
made off-line programming a realistic of NC machines. Thus robot tasks are An additional feature of many robot
proposition for groups of more complex conventionally programmed at the simulators is the availability of postpro-
machines indirectly communicatingwith machine controller. A range of pro- cessing software for off-line program-
each other through shopfloor computer prietary programming languages have ming of robots. Such a postprocessor
systems. Unfortunately, the design and become available for this task - for reformats the geometric and sequential
programmingof a manufacturingsystem example VAL I I , AML, KARREL and HELP information generated by the modeller
which includes industrial robots can (Refs. 1 and 2) - and are used in con- and simulator to produce a robot task
involve far greater complexity than that junction with a teach pendant. While program in the native language of the
for a system composed solely of NC robot languages of this type simplify robot. Obviously this postprocessing
machines. This i s a direct consequence task programming, they have attendant function is robot dependent (currently
of the wider variety of possible applica- major disadvantages, as follows: there is no internationally accepted
tion areas and potential solutions given neutral language for robots), and hence
the enhanced flexibility offered by 0 For small-batch production, loss of this facility will only be readily available
robotics. production time in the teach mode (while for commonly used industrial robots.
The designer of a robot system must defining the workplace geometry) can Let us consider some of the necessary
determine an appropriate choice of be significant. features of the modeller. Commonly
robot and its workplace elements, inclu- 0 It i s necessary to recreate infor- either ’set-theoretic’ (Ref. 7 ) or ‘bound-
ding feeders, tools, grippers, fixtures and mation (concerning processes and pro- ary representation’ (Ref. 8 ) solid model-
sensors. The selection of a particular ducts) which may already be available in ling techniques are used to allow a
solution from the very many valid a machine-readable form elsewhere in three-dimensional.model of the robot
alternatives must be determined by the manufacturing system. and workplace to be constructed from
Computer-Aided Engineering Journal August 1988 157
simple primitive shapes such as cuboids, represent the movement of the robot volved in attempting a full dynamic
regular prisms and cylinders, or gener- (relative to its peripheral equipment, simulation would be justified except in
ally by closed polyhedra. Typical robot namely machines, feeders, conveyors rare circumstances. However, facilities
simulators will facilitate model creation and components etc.) between these for obtaining cycle time estimations can
by ’easy-to-use’ data input methods, and various workplace co-ordinate positions. be particularly useful as this information
by allowing the created models to be Path control can be imposed on the can be used in investment appraisal and
stored in and recalled from a library. motion between positions so that resource planning exercises.
When created, the robot and workplace straight-line or circular trajectories can
model may be viewed and manipulated be represented. During this simulation Limitationsof the present generation
in standard ways, for example displayed process, interference detection software of robot simulators
graphically in plan, front and side ele- can be executed so that collisions in the
vations, and in perspective and plane workplace can be flagged. This is in- Here we consider three key functional
parallel projections from any viewpoint. herently possible through the use of set- areas in which the present generation
The spatial relationship between entities theoretic or boundary representation of robot simulators demonstrate limi-
in the model (the robot and its equip- modelling techniques, although it can tations, these limitations reducing their
ment, for example) can be controlled be extremely time consuming. effectiveness and widespread usage.
using the normal CAD input and display The present generation of robot simu- These functional areas are ’calibration’,
practices. lators can be considered essentially to ‘integration’ and ’ease of use’.
Typically methods are also provided to be kinematic in nature, although com-
allow robot model movements to be monly an estimate of the cycle times Calibration
controlled in conventional ways, such as related to various workplace movements In common with any simulation pro-
in ’world’, ’tool’ and ‘joint‘ co-ordinate can be obtained: velocity information cess, the usefulness of a robot simulator
systems. For example, the user may being assigned to the manipulator is governed by the‘accuracy’with which
specify the desired position of the tool model at model creation. Thus the the robot and its workplace can be
centre point, with the robot simulator dynamic characteristics of the robot and model led.
performing the inverse transformations its workplace elements are not usually Using the present generation of
necessary in order to determine and accurately modelled. For example, in a computer systems, sufficient processing
display the robot posture. A number of high-speed contouring application the power can be made available to achieve
such positions can be stored as a series real manipulator will be subject to very high precision modelling of any
of sequential events in a manner anal- backlash , deflect ion of manipulator manipulator system. However, in cre-
ogous to the teach method of robot links, following error etc. We shall ating such a model, specific input data
programming. Subsequently, a time- return to this point later, but it is concerning the manipulator’s kinematic
based simulation can be executed to debatable whether the complexity in- and dynamic behaviour must be available

RedBoard PCB

SFC update
modules

SFC C A D
pre-processor updated
locations

download
v i a RS 232
serial link
1 , seriol link
I
workplace
equipment
robot
sequence board robot
model
1- controller
h-’
1 --l
model model
robot

L robot
i

Fig. 1 Operational schematic diagram of the project software

158 Computer-Aided Engineering Journal August 1988

from some source. When simulating
robots, there are two obvious sources of
input data:

0 from the robot manufacturer

through using manipulator and work-
place measurement devices.

If the robot simulator i s used only in

choosing a robot and designing a suit-
able workplace layout the first of these
sources can be appropriate; and this
approach is favoured by current sup-
pliers of robot simulators. However,
robot manufacturers usually supply
only limited statistically averaged kine-
matic data concerning their manipu-
lator, with often little or no data relating
to its dynamic behaviour. Thus when
compared with simulated results a
specific industrial robot will demon-
strate both dimensional variations (result-
ing from manufacturing tolerances, back-
lash, deflection in manipulator links,
control system resolution and dead-
band etc.) and unpredicted dynamic
behaviour (damping, following error
etc.) as it is moved through its working
envelope. Clearly, any significant in-
accuracies or omissions in specifying
the robot and workplace model can lead Fig. 2 Typical simulation programming exercise
to lack of confidence in the simulation
results. 0 through the use of workplace sensors components into a family of printed
This situation is further exacerbated to calibrate the robot and its workplace. circuit boards (PCBs). Suppose also that
when attempting to use the simulator the PCBs are to be manufactured in
to achieve the off-line generation of The use of a teach pendant to overcome small batches and have been designed
robot task programs. Here it is not modelling errors must be viewed as a using a proprietary CAD facility. In such
sufficient for workplace elements to retrograde step, unless the pendant i s an instance, various information con-
assume nominal positions, shapes and, used only to establish a limited number cerning the artwork on the PCBs, the
where appropriate, dynamic behaviour. of reference or datum points, while component types and their geometry,
Nor can significantly large dimensional the use of workplace sensors will incur and the required location of the compo-
and dynamic behavioural errors in the a significant increase in complexity and nents on the PCBs will all be stored
manipulator simulation be tolerated. additional cost. In fact, generic robot within libraries in the PCB design system.
The capabilities of current solid model- calibration procedures have yet to Although this information is likely to
ling techniques must be questioned for evolve in an internationally accepted be stored using proprietary data formats,
certain applications of off-line program- sense. This can be attributed to the it will exist in a machine-readable form
ming. The representation of geometry complexity involved in accurately and can be utilised in generating the
by plane or simply curved surfaces may measuring the position and orientation workplace model. Clearly, however, a
be adequate for simulation purposes, of workplace elements, and their dy- data link of some form (which may be
but could be too gross an approxi- namic characteristics. However, the automated or involve manual interven-
mation where the parts being manufac- McDonnell Douglas Robotics Software tion) can be established and processing
tured are composed of complex curved suite includes an ’ADJUST’module which facilities provided to re-format the
surfaces. Although hybrid CAD model- offers a specific solution to this problem information into data structures which
ling systems are under development, through the use of a calibration probe, can be integrated within the workplace
current set theoretic and boundary mounted on the robot’s face plate or in simulation model. Such an arrangement
representation solid modellers lack the the gripper (the probe supplying could yield significant time savings in
complex surface representations to be feedbackhnput data to the robotlwork- any robot/workplace simulation and off-
found in surface modelling systems. place model). line programming venture.
To date very little detailed technical This specific example serves to
information i s published in the litera- Integra tion illustrate that for robot simulators the
ture which documents the use of robot As indicated earlier, future genera- modelling process can be considered to
simulators in off-line programming ap- tions of robot simulator may have an comprise three constituent elements,
plications. However, two approaches increasingly important role in the evolu- namely models of:
have been used in attempting to over- tion of CIM systems. Here we will con-
come modelling errors: the robot manipulator
sider the possible form of an interfaceto
the robot workplace and its tools
product design.
the product.
0 through two-stage programming, Consider the specific case of an elec-
the second involving the use of a teach tronic manufacturing environment where Having chosen a robot and determined
pendant a robot is to be used to insert ’odd-form’ its attendant workplace equipment, the

Computer-Aided Engineering Journal August 1988 159

 can only suggest that the user interface
to a robot simulator should be as simple
as possible and should make extensive
use of model libraries, whereby not only
the robot but also commonly used
workplace elements can be easily con-
structed through parametrising standard
primitive building blocks. Furthermore,
where a product design interface cannot
be established, facilities could be in-
cluded for the user-friendly description
of products encompassing geometric
Fig. 3 Typical pallets
and sequential information.
As a generalisation one could conclude
that off-line programming will be more
easily justified in circumstances where
complex robot tasks are required and/or
where the batch manufacture of large
product families is involved. However, at
the moment inadequacies in the simu-
lators' capabilities in dealing with sensory
feedback, exception handling and the
debugging of task programs imply the
continued use of on-line programming
for these aspects, where a language
such as VAL I I combined with a teach
pendant have the necessary flexibility.

A research study at Loughborough

University of Technology

A research study at Loughborough Uni-

versity of Technology has involved an
appraisal of robot simulation and off-
line programming methods. Thus far the
research work has progressed along
three themes, with software being pro-
Fig. 4 Example workplace configured using the model building facility duced to facilitate calibration, an inter-
face to product design and improved
arrangement so formed will remain fixed the success of standards initiatives such man-machine interface (MMI) capabiIi-
in most situations, and so too will the as MAP/TOP (Ref. 9), EDlF (Ref. IO), ties. This software has been designed
associated model elements in any simu- ICES (Ref. 11) and PDES (Ref. 12). Such specifically to operate with the GRASP
lation. However, when batch manu- specifications will allow manufacturers robot simulator, but future work will be
facturing is involved, changes in the to supply their automation products aimed at generalising the functionality
product occur. Thus where a large with standard interfaces, thereby allow- involved.
family of products are involved, or ing standard product descriptions to be Fig. 1 illustrates conceptually the
where complex geometric descriptions stored and transmitted across those overall structure of the functionality in-
of products are concerned, a link to pro- interfaces. cluded so far, where the off-line pro-
duct design can be of significant benefit. gramming of two AdeptOne SCARA
Furthermore, significant benefit can 'Ease of use' robots, tooled-up for PCB assembly, has
be gained by establishing data links to The premise on which off-line pro- been accomplished. Fig. 2 illustratesthis
other computer-based manufacturing gramming systems have evolved and are activity by showing a view of a specific
activity areas. For example, an interface marketed i s that they are easy to use. simulation exercise.
to an automated process planning system Certain suppliers of robot simulators Although progress has been made in
could access valuable information con- claim that their system is user-friendly, relation to all three themes of study, in
cerning the sequence of manufacturing so that most operators, with no previous this paper we only describe the project
operations to be performed. In our PCB programming experience, can design activity in relation to enhancing the user
assembly example, process planning workcells after one day's training. Those interface to GRASP.
information might already exist, in suppliers have also claimed that an As discussed earlier, the MMI to
machine-readable form, describing the operator can become an expert in two proprietary robot simulators can be con-
sequence in which components should to four weeks compared with six to nine sidered within two major groups of input
be inserted into the PCBs. Thus a data months for a general-purpose CAD facilities: one being related to the solid
link could facilitate the input of infor- system. This may in fact be the case in modeller, with the other related to robot
mation, defining theorder in which sub- simple application areas, but to date task description.
tasks should be performed to the simu- very little literature i s available com- The solid modelling facility within
lation process and subsequently to the paring off-line and conventional robot GRASP is based on the boundary repre-
robot task program. language programming procedures so sentation method, using a combination
The success of any integrated solu- that reliable conclusions can be drawn. of primitive solids and generalised poly-
tions as described will rely heavily on In the absence of information, one hedra. A robotic device or workplace

160 Computer-Aided Engineering Journal August 1988

entity is modelled by defining geometric, modelling of grippers, assembly jigs the pallet dimensions and name, are
spatial and functional relationships. Defi- etc. can be extremely difficult, and a supplied in response to prompts, and
nition of geometry using primitive shapes parametric approach could provide a are used to generate a textual GRASP
is straightforward, but generalised poly- solution. source file. The GRASP pictorial repre-
hedra present a much more substantial Solid modelling software modules sentation of typical pallets is illustrated
problem which requires the assistance have been coded in Pascal, and these by Fig. 3.
of screen interaction techniques. Spatial can assist both experienced and inex- The same approach, of generating
relationships between objects in three- perienced programmers when creating workplace entities based on user
dimensional space inevitably cause dif- models of entities commonly encoun- responses to simple input prompts, was
ficulties which can be addressed by an tered in a robot workplace, for example used in creating other commonplace
improved user interface or by improved, pallets (or racks), worktables (or benches), workplace entities. Software modules
but potentially expensive, computer conveyors, turntables, tool magazines were also produced for tool magazines
graphics techniques. Functional arrange- etc. Here the programmer is required to and work surfaces, as well as kinemati-
ment refers to the need to encapsulate input a limited set of parameters so that cally functional turntable and conveyor
mechanical and kinematic relationships a text source file can be generated auto- entities. Fig. 4 illustrates workplace
between model items within the data matically and coded appropriately for models created in this way, where ap-
specification. For inexperienced users input to the GRASP solid modeller. The propriate geometric, spatial and func-
or complex situations this can be an software modules produced have a tional relationships have been defined.
extremely daunting prospect. common design structure. Clearly, the time taken by a user in
Once the model has been created To illustrate the principles involved let creating a model depends upon the
there still remains the often complex us consider a specific software module complexity involved and hidher level of
task of building the kinematic simulation which was produced to assist users in expertise. For the examples treated here,
itself. creating objects in the pallet or rack work surface modelling is the simplest,
Thus methods of enhancing the user category. The parameter values, such as this also being reflected in the level of
interface to GRASP have been studied
and software modules have been pro-
duced to assist both experienced and
non-expert programmers in dealing with
solid modelling and task programming.
(7 start

Computer-assisted model building

Building a three-dimensional model is
a requirement which robot simulators
have in common with CAD systems
aimed at the more traditional aspects of
e= input track name

before? instruction /
engineering design. Simulators that are
an extension of such systems are there-
fore likely to be able to call upon highly
developed MMls for the specification L
I
display
yes
I
I
of geometry. A typical approach would options
be a parametric design programming
language often used to ease the gen-
eration of models of families of parts.
GRASP, a stand-alone simulator with its
own modelling system, does not have
such facilities, and thus provides both
the need and the opportunity to study
parametric design from the particular
viewpoint of robotic equipment.
I 1
The distinguishing feature here is the
need to define easily the functional re-
lationships between the components of
the equipment. An example would be in i f options with
specifying geometry of a gripper such pattern defined by
that it will have the correct relationship reference points
to the robot on which it is to be
mounted. (In the case of GRASP, this
involves careful modelling about a set of
axes which will be made coincident with
those of the tool attachment point.)
Careful modelling is again required
when the object being modelled is a
mechanism which will be required to I U
assume different configurations under
various kinematic conditions. Modelling
the kinematic structure of the robots
themselves is of course a central aspect
of the simulator and hence is well
catered for. However, the kinematic Fig. 5 Module structure used to define palletising operations

Cornputer-Aided Engineering Journal August 1988 161

complexity in the associated software object, so as to be able to grip it and lift and/or where batch manufacturing of
module, whereas conveyor modelling is it up before the remainder of the task large product families is involved.
the most complex. When using the continues. if off-line programming is to be
modelling facilities of a typical robot Clearly there are many other ‘micro implemented using graphical simu-
simulator a rough estimate of the time sequences’ of this type, which when lators of the type described in this
needed to create these solid models is defined and parametrised would make paper, then a number of shortcomings
between 10 and 30 minutes, the actual extremely useful building blocks for within these systems must be overcome.
time taken being dependent on the ‘macro sequences’ or complete robot Thus any model is an approximation to
experience of the user (although a tasks. the real situation, and its use to deter-
significant part of the time required is To demonstrate this principle a soft- mine the explicit activities of the real
usually spent on calculating the correct ware module was produced to simplify robot working in its real environment
geometric location of the primitives). the definition of a variety of palletising must result in a requirement to calibrate
However, if the enhanced MMI software operations. This module comprises a the model. The proposed method of
modules are used in generating an number of functional elements, as shown calibration would involve workplace
entity model, then the time required schematically in Fig. 5. sensors feeding back deviation infor-
will be between 1 and 2 minutes. Thus In simulating a task it is necessary to mation to the model, thus permitting
significant time saving can be achieved. define any reference points, and these the generation of an accurate program.
Perhaps of greater significance, how- may take the form of an array. To simplify This leads directly to the second limit-
ever, is the confidence that can be reference point definition, optional soft- ation: inadequate integration with design
placed in a parametrically defined model ware-based procedures were written and manufacturing control systems. The
object. The parametrisation will have which accept the names of reference solution here would seem to be the use
been subjected to vigorous testing of its points. The resulting simulation task is of emerging standards for both manu-
suitability for the application, and thus presented in GRASP syntax, ready to be facturing information and the method
it is unlikely that model shortcomings used in simulation. Subsequentlyduring of its transmission between design and
will be discovered inconveniently late in simulation any joint violation and object production facilities.
the simulation process. collisions can be detected. The computer- The final area of concern is the ease of
At present, a limited range of items assisted model building and task oper- use of the simulation systems. Designers
can be produced through the use of ation facilities have been used in pro- and robot programmers are being asked
specific software modules, but the dis- gramming a number of tasks for two to perform an extremely complex ab-
cussion above illustrates the general AdeptOne robots, the off-line program stract task, the results of which largely
principles involved. However, it is recog- generat ion being accom pii shed through determine the efficiency with which
nised that eventual full-scale implemen- the use of a VAL I1 postprocessor. These expensive manufacturing facilities are
tation is likely to be via a general-purpose off-line programming exercises have utilised. To do so they should be pro-
parametric languagetaking into account proven the validity of the approach and vided with tools which are not only
the special needs of this technique in suggested many possible enhancements. functionaily satisfactory, but also meet
robot modelling. high ergonomic standards.
It can been seen then that off-line
Conclusions
Computer-aided task program programming by graphical computer
generation of simulated tasks methods has considerable potential
GRASP does have a language for the Off-line robot programming will be advantages, but these are tempered by
definition of a sequence of events more easily justified in circumstances a consideration of some fundamental
within the simulation. This is very flex- where complex robot tasks are required limitations of existing systems.
ible, being accessed through screen
menus or by a textual command
language, and can be used to explicitly
define a wide range of tasks. It is, References
however, still a complex task to create
new sequences in a generalised way, 1 LOZANO-PEREZ,T.:, ’Robot programming’, Proceedings of IEEE, 1983, 71, (7), pp. 821-841
and therefore software modules that 2 BONNER, S., and SHIN, K.: ’Comparative study of robot languages’, Computer, 1982,
can simplify this activity have also been pp. 87-97
produced. 3 YONC, Y.F., CLEAVE, J.A.,GREEN, J.L., and BONNEY, M.C.: ’Off-line programming of robots’,
In particular, common material in ’Industrial handbook on robotics’ (Wiley, 1985)
handling tasks have been studied and 4 STAUFFER, R.N.: ‘Robot system simulation’, Robotics Today, 1984, 6, (3), pp. 81-90
5 HOWIE, P.: ’Graphic simulation for off-line robot programming’, Robotics Today,1984,6, (I),
software modules have been produced
pp. 63-66
for generating palletising, de-palletising
6 EVERSHEIM, M., WECK, M., SCHOLINC, H., SUHLKE, D., and MULLER, W.: ’Off-line
and machine tending tasks, which are programming of numerically controlled industrial robots using the Robex-programming
representative of tasks for which robots system’, CIRP Annals, 1981, 30, (I),pp. 419422
are frequently used. Here the command 7 REQUICHA, A.A.G.: ’Representations for rigid solids: theory, methods and systems’,
inputs invoke commonly occurring se- Computing Surveys, 1980, 12, (4), pp. 437464
quences of motion when performing 8 Research Reports, ESPRIT Project 322: ‘CAD interfaces (CAD*I)’.Vol. 1, Version 2.1
simulations. For example, it is not usual 9 HOLLINGHAM, J.: ’The MAP report’ (IFS Publications, 1986)
practice to pick up an object directly as 10 HILLAWI, J.I., and BENNETT, K.R.: ’EDIF - an overview’, Computer-Aided Engineering
collisions between the manipulator’s Journal, 1986, 3, (3), pp. 102-107
11 WILKINSON, D., and HALLAM, R.: A ‘ study of product data transfer using ICES’,Computer-
end-effector and other workplace entities
Aided Engineering Journal, 1987, 4, (3), pp. 131-136
may occur. Instead, the manipulator i s 12 SMITH, B.M.: ‘A reporting of the PDES initiation activities’. National Bureau of Standards,
normally commanded to move to a Caithersburg, MD, USA, 1987
position above the object (if this is clear
of obstacles), from which the gripper i s S. F. Chan, Prof. R. H. Weston and Dr. K. Case are with Department of Manufacturing Engin-
moved in a controlled fashion on to the eering, Loughborough University of Technology, Loughborough, Leics. LEI1 3TU, England.

162 Computer-Aided Engineering Journal August 1988