robots more accurecy more precise ) CLASSIFICATION OF ROBOTS AND THIER GEOMETRIES )

Definition of a Robot

According to The Robot Institute of America (1979) : "A reprogrammable, multifunctional manipulator designed to move materials, parts, tools, or specialized devices through various programmed motions for the performance of a variety of tasks."
According to the Webster dictionary: "An automatic device that performs functions normally ascribed to humans or a machine in the form of a human (Webster, 1993)."

II. History of Real-World Robots:

One of the first robots was the clepsydra or water clock, which was made in 250 B.C. It was created by Ctesibius of Alexandria, a Greek physicist and inventor. The earliest remote control vehicles were built by Nikola Tesla in the 1890's. Tesla is best known as the inventor of AC electric power, radio (before Marconi), induction motors, Tesla coils, and other electrical devices. Other early robots (1940's - 50's) were Grey Walter's "Elsie the tortoise" ("Machina speculatrix") and the Johns Hopkins "beast." "Shakey" was a small unstable box on wheels that used memory and logical reasoning to solve problems and navigate in its environment. It was developed by the Stanford Research Institute (SRI) in Palo Alto, California in the 1960s. The General Electric Walking Truck was a large (3,000 pounds) four legged robot that could walk up to four miles a hour. The walking truck was the first legged vehicle with a computer-brain, developed by Ralph Moser at General Electric Corp. in the 1960s. The first modern industrial robots were probably the "Unimates", created by George Devol and Joe Engleberger in the 1950's and 60's. Engleberger started the first robotics company, called "Unimation", and has been called the "father of robotics."

III. Modern uses of Robots:

1) for EXPLORATION

People are interested in places that are sometimes full of danger, like outer space, or the deep ocean. But when they can not go there themselves, they make robots that can go there. The robots are able to carry cameras and other instruments so that they can collect information and send it back to their human operators.

The "Odyssey IIb" submersible robot is shown suspended in a tank. It was developed by research scientists at M.I.T. for ocean exploration. The inset shows the "Sojourner" microrover robot being repaired at the Jet Propulsion Labs. Sojourner landed on the surface of Mars on July 4, 1998 (from National Geographic, July 1997).

2) for INDUSTRY

When doing a job, robots can do many things faster than humans. Robots do not need to be paid, eat, drink, or go to the bathroom like people. They can do repetative work that is absolutely boring to people and they will not stop, slow down, or fall to sleep like a human.

"Industrial robots spot weld automobile bodies on an assembly line" (from National Geographic, July 1997).

3) for MEDICINE

Sometimes when operating, doctors have to use a robot instead. A human would not be able to make a hole exactly one 100th of a inch wide and long. When making medicines, robots can do the job much faster and more accurately than a human can. Also, a robot can be more delicate than a human.

"ROBODOC", a modified industrial robot, drills a precise hole in the femur (thigh bone) of this skeleton (from National Geographic, July 1997).

Some doctors and engineers are also developing prosthetic (bionic) limbs that use robotic mechanisms. Dr. David Gow, of the Prosthetics Research and Development Team at Princess Margaret Rose Orthopaedic Hospital, made the first bionic arm called the Edinburgh Modular Arm System (EMAS) in 1998. See also the news article from BBC NEWS ONLINE and the bionic arm web page of the Princess Margaret Rose Orthopaedic Hospital.

Campbell Aird, Scottish hotel owner, fitted with the world's first bionic arm (images from The Irish Times web site).

4) for the MILITARY and POLICE

Police need certain types of robots for bomb-disposal and for bringing video cameras and microphones into dangerous areas, where a human policeman might get hurt or killed. The military also uses robots for (1) locating and destroying mines on land and in water, (2) entering enemy bases to gather information, and (3) spying on enemy troops.

5) for ENTERTAINMENT

At first, robots where just for entertainment, but as better technology became available, real robots were created. Many robots are still seen on T.V. (Star Trek - The Next Generation) and in the movies (The Day the Earth Stood Still, Forbidden Planet, Lost in Space, Blade Runner, Star Wars). These imaginary robots do alot of things that the real ones can not do. Some robots in movies are made to attack people, but in real life they cannot really hurt people at all because they are not in control of themselves. Robots also attack humans in video and computer games. So don't think all robots do is kill, because they can't.

A Robot Controller Using Learning by Imitation

Abstract.

Roboticists have already invested considerable energy in building robot controllers which model the learning capacities of single animals In this paper we present a new type of controller which draws upon insight from he eld of imitative learning one agent learns from perceiving and imitating the behaviour of another We describe the architecture of an imitative learning controller and two implementations a simulator and a robot The learner robot follows a teacher robot through a maze and learns to associate its perceptions at locations where the teacher carries out a signi
cant action with the action it subsequently undertakes as a result of its innate teacher following behaviour Such a controller limits the learning task to bouts of learning when there is something useful to be learnt It allows a robot to learn in terms of its own perceptions makes programming many nominally identical robots simpler and opens the possibilities for cross modal learning.

INTRODUCTION.

Roboticists have invested considerable eort in implementing biologically inspired control mechanisms in robots They have made progress in the design of reex behaviour for robots and have started to make inroads into the eld of robot learning for example by programming robots to form associative maps of their environment or to learn from reinforcement received from the environment for successful behaviour In this paper we present a controller which utilises a new type of learning imitative learning whereby one robot learns how to act by perceiving and in some sense imitating the actions of another robot Once it has learned it is able to negotiate new situations on its own Imitative learning is a term taken from ethology which ethologists use to describe a huge range of behaviour in animals Indeed there is substantial disagreement among ethologists as to what exactly constitutes imitation in animals since the term is used to describe phenomena as disparate as vocal mimicry in birds the tendency of chicks to eat more when in the company of conspecics and faster learning of tasks by rats or kittens who are able to observe ateacher The mechanisms postulated by ethologists to underlie these phenomena are correspondingly disparate Zajoncsuggests that the mere presence of conspecics may make it more likely thatan animal carry out a particular behaviour on the other hand it may benecessary for the conspecics also to be undertaking the same behaviour for facilitation to occur Simple fear reduction in the presence of others may be taking place so that innate behaviours are disinhibited rather than being positively activated These mechanisms could explain such social facilitation behaviour as the tendency of chicks to peck at food more when others are around Thorpe suggests that contagious behaviour may result if the sight of one animal performing some innate behaviour is the stimulus which triggers this behaviour in another as in for example yawning in humans or ock manoeuvres in birds On the other hand Humphrey suggests that classical conditioning could be involved if some stimulus always releases the same behaviour in a group of animals then that behaviour in one animal is the conditioned stimulus for the same behaviour in another Yet more mechanisms are postulated to underlie other sorts of imitative behaviour In general the specications of these mechanisms given in ethological descriptions are not sucient or not in the right form to be directly implementable as a robot controller However such a controller would clearly make robot programming easier Even robots which are nominally identical cannot be made to behave robustly by the simple expedient of copying a program from one to the next no two robots are identical and no two sensors the same We therefore adopt the following strategy. First we allow a very general denition of imitation as being a phenomenon which leads to some sort of similarity in behavior among two or more individuals We then pick a mechanism put forward as a possible explanation of one type of imitative learning matched dependent behaviour and design and build a robot control system based on this Our main aim is the construction of a robot controller that is robust and versatile so that our system and the expressed behaviour of the robot may well not be completely faithful to the ethological case Finally our experiences with this system may allow us in the future to further our second aim the elucidation of the ethological mechanisms which are hypothesised to underlie particular types of imitative learning.

Matched Dependent Behaviour.

We take matched dependent behaviour to refer to the phenomenon whereby for example rats negotiating a maze receive reinforcement faster if they match their behaviour to that of a leader rat the leader knows how to nd the food and then learn to associate the other stimuli in the environment at the time presumably .
those which stimulated the leader to do the right thing in the rst place with the appropriate action They are then able to carry out that action in the absence of the leader In our system we concentrate particularly on the second stage of this process A teacher robot negotiates its environment a learner robot follows and watches for signicant events in the behaviour of the teacher The learner matches its behaviour to the teachers and learns to associate the stimuli in the environment at the points at which the teacher performs signicant actions with the behaviour it subsequently carries out by virtue of imitating the teacher Currently no reinforcement is obtained by either the teacher or the learner The teacher is able to negotiate its environment without reward this behaviour being simply programmed in and the learner has no choice but to follow the teacher Since our initial aim is to build a useful robot controller omission of the generally slow computationally expensive reinforcement learning stage is permissible it should of course be included if we intend to provide an explanation of the biological phenomenon In robotic imitative learning perception is clearly crucial The learner robot must be able to perceive what the teacher robot is doing and to respond preferentially to the relevant stimuli in its environment Perception is usually dicult there is the conceptual problem of deciding just which stimuli are the relevant ones and there are often the practical problems of accommodating on board the robot the large amounts of computational power needed for example if complete video images need to be processed to a high level of abstraction and of getting this processing done in real time Whilst not denying these diculties we contend that useful imitative learning can be carried out using simpler detectors such as infrared distance sensors and simple patterns of movement such as straight line locomotion and deviation therefrom and we have chosen to use these in the rst stage of our investigation of the architecture of a complete controller We have opted initially to have the teacher robot travel through a maze negotiating corners when it reaches them with the learner following a certain distance behind The learner notices when the teacher Changes its direction of travel a signicant event in order to negotiate a corner The learner then associates what it perceives in the environment around it with the action the teacher has carried out and imitates that action Eventually the learner is able to make use of these associations to respond to corners correctly on its own Our choice of experimental testbed was governed by the need to be able to evaluate whether the robot had actually learned anything By choosing to teach a robot to navigate a maze and then testing its learning by asking it to negotiate either the same or a new dierentaze we have ademonstration that learning has taken place.

Architecture.

In this section we describe the general architecture which governs our realisation of the matched dependent behaviour mechanism The framework is very general although we have described it as it is used in maze negotiation Each module requires dificult aspects to be addressed there will be many possible solutions often research problems in their own right We go on to describe two implementations of the architecture in simulation and on autonomous mobile robots and present our rst results The architecture comprises ve modules which we describe in turn Maze negotiation The teachers maze negotiation behaviour is implemented in the rst module It must allow the teacher to respond appropriately to situations such as confronting a wall right and left corners Tjunctions and so on Its implementation could take a very simple form such asdead reckoning or one could use sensors and a negotiation strategy A reinforcement learning module could be useful although taking many trials to learn it would be expected to be insensitive to sensor noise and small variations in the environment Alternatively the robot could be led through the maze several times by a human and learn using the fast Halperin net which takes advantage of designer knowledge of the task and so does not need to learn the environment and task from scratch In the general case this module would be responsible for producing whatever behaviour the teacher is teaching there is no limit to the number of dierent instantiations and complexity of the module Teacher following In this module which is independent of the maze negotiation module the learner robot is programmed to follow the teacher In the simplest implementation the learner keeps axed distance behind the teacher regardless of the teachers motion more sophisticated implementations could have the learner follow only when the teacher is moving purposively and ignore loitering In both cases the learner must be able to detect the teacher and its distance from the teacher and have strategies to deal with the situation when the teacher moves out of sight In the general case this module is responsible for making the learner imitate the behaviour of the teacher When the teachers behaviour can be described as bulk motion the implementation of the module is quite straightforward however how the learner might imitate the teachers behaviour in say a manipulation task involving arm movements is less clear The learner would then need some understanding of what the teacher is doing from the teachers point of view in order to be able to match its behaviour to the teachers In the case of simple motion it is much easier to insulate what the learner knows from what the teacher knows with the sole means of communication between the two being the perception by the learner of the teachers behaviour SignIcant event perception The learner robot recognises when the teacher robot is performing an action it deems signicant and remembers the place at which that action is being performed Ways of recognising that an action is signicant include detecting a change in direction of movement or detecting a change in the direction in which the teacher is pointing the teacher could also signify it is about to undertake a signicant action by emitting a sound watch me This module allows the learner to recognise that one of the actions it is about to do we assume that the learners behaviour time lags the teachers is to be distinguished from the rest from this point onwards the learner must be ready to learn Thus the learner engages in short bouts of learning which coincide with those periods in which there is something useful to be learnt it does not have to process a continuous stream of perceptions and associate them with a continuous sequence of actions By directing the learners attention the complexity of the learning process is reduced Association environment perception The learner robot knows from the signicant event perception module the location of and its distance from the signicant event It moves to this location and then perceives its environment This is the stimulus with which an appropriate action is to be associated The implementation depends very much on the sensing capabilities of the learner which could vary from simple binary distance detectors to general purpose vision systems In most cases one would expect to use specialised sensors tailored to the environment The learner must be able to discriminate suciently the salient features of its environment but it must also be able to generalise well enough that when it self in a situation which is similar to but not identical with the rst it is still able to carry out the appropriate action Association recognition of own action Finally the robot must associate an action with the stimulus it has just received from the environment this is essentially the imitated action In our systems we make use of the fact that the learner is performing a teacheR following behaviour all the time thus in order to discover the appropriate action the learner simply has to move a few steps further in its path behind the teacher and keep note of what exactly it does and it will automatically have carried out the correct action The sequence of learner movements just before and just after the signicant event is the action that is associated with the perception The implementation of this module could vary widely and there is also room for variation in the design of the action recognition itself The form of the association itself can be simple or complex from a set of if then rules to a connectionist network We are currently using rules although further connectionist developments are envisaged once the amount of information provided by the sensors is large enough The complexity required depends in large part on the variability in the environment the noise in the sensors the precision with which the learner can measure its own actions and the degree to which a robots behaviour can be characterised as a sequence of signicant recognisable events.

Results from Simulation.

We have implemented this architecture on a simulator which was built to allow us to rapidly test the applicability of this and other architectures The current implementation on real robots is described below The implementation of learning on the simulator has two phases Learning phase Robot teacher executes a series of moves which take it through the maze and the robot learner follows the steps of the teacher associating environment congurations with actions followed when something signicant happens Testing phase The robot learner equipped with the rules it acquires from the learning phase is requested to traverse either the same or a new dierent maze on its own During therst phase Figure a series of moves is executed by the robot teacher one of MoveForward MoveLeft MoveRight while the robot learner follows the teacher by maintaining a close distance and changing direction with a suitable time lag whenever the teacher does so The teacher is never allowed out of sight of the learner The orientation of the teacher is detectable by the learner a change in the orientation by degrees of the teacher is the signal for the learner that something signicant has happened The learner waits until it reaches the point of change since it knows how far behind the teacher it is it just has to travel this distance towards the teacher Then it invokes the association procedure This involves sensing of the environment around it ie the conguration of the walls essentially the type of corner it is dealing with and associating that with the appropriate action In order to determine this action the robot waits to see what it des next governed by the instinctive following behaviour It calculates the action as the dierence between the direction in which itwas travelling before it reached the signicant event location and the direction in which it is travelling afterwards In our particular implementation this is either a rotation of degrees clockwise or degrees anticlockwise After the learning phase we can see that these turns are associated with the right and left corners respectively During the second phase Figure the robot learner is placed at the beginning of either the same or a dierent maze and is requested to traverse it without the help of the teacher It constantly senses the environment around it attempting to match the conguration of the walls with those specied as signicant congurations in the left hand side of the rules acquired during the learning phase When a particular conguration matches the operator which is associated with it is applied to the robot causing a change of direction and then the robot continues moving forward again The results so far are encouraging In the simulated implementation the robot learner successfully learns to traverse increasingly complicated mazes An example of the rules acquired is shown in Figure where the default learner behaviour is also shown as a rule The architecture is capable of dealing with simple learning situations Our next step with the simulator will be to increase the complexity of the maze negotiation and teacher following modules and particularly the form of the association modules It is unrealistic to expect to be able to simulate the robots sensing capabilities properly for this we have turned to real robots.

Robot Implementation.

In this section we describe how the imitative learning system is implemented to control the behaviour of two robots A Lego robot of the type used for research and teaching in our Department is used as the robot teacher This is a small carlike vehicle with two motors powered by batteries augmented with abased microprocessor board Brain Brick and various specially built sensing bricks which process the output of infrared light dependent resistor Hall e ect and bump switch sensors One of the Departmental research robots Ben Hope serves as the learner This is an autonomous transputer based mobile robot cylindrical in cross section and about cm high It is equipped with a video camera a bumper and motion detection ultrasonic and infrared sensors it has a radio modem for communication with a host PC if required All processing takes place on board the robots although data can be stored and downloaded to a workstation Our initial implementation has the Lego robot navigating through a maze of wooden boards about cm high Currently we are most interested in experimenting with the learning modules so we have chosen to have the teacher navigate by dead reckoning A smarter strategy will be used when we switch the environment to a more complex corridor and oce area The learner Ben Hope follows using its camera to sense a pointing device mounted on top of the Lego robot This device is a triangular conguration consisting of ve LEDs positioned in a row with another single LED situated about cm from the row The LEDs are visible in a video image from Ben Hopes camera making it a simple matter to detect both the Lego robot itself and the direction in which it is moving The view of the teacher seen by Ben Hope is shown in Figure Ben Hopes camera has quite a wide eld of view so we discard the upper part of the image before thresholding it and applying Haralicks iterated components algorithm see to extract connected labelled areas Figure The positions of the front and back of the teacher robot and the direction in which it is pointing can then be obtained by tting ellipses to the two blobs Given the height above ground of Ben Hopes camera it is simple to calculate the distance to the Lego robot All of the image processing is carried out on Ben Hopes framegrabber TRAM The speed of Ben Hope is then controlled to keep it axed distance behind the teacher robot eectively we are trying to keep the position of the centroid of the teachers pointing device within a small region round the centre of the image The direction in which Ben Hope is travelling can also be adjusted so that it follows the teacher round corners A signicant event is signalled by the orientation of the teacher robots pointing device changing by some threshold amount Maze wall con gurations are detected by the infrared sensors mounted around Ben Hope and the action to be associated with this perception is the degree of turning required to bring the teacher robot back into Ben Hope seld of view As in the simulator the acquired association of perception and action takes the form of simple rules Work is currently being carried out on the integration of the semodules we hope to present the results of the learning experiments at the symposium.

Conclusions and Summary.

We have described the architecture we are using to investigate a new mechanism of robotic learning namely imitative learning and two implementations applied to a maze problem one in simulation and one using robots Results from the simulation are promising the robot system is being integrated The use of this particular architecture has certain advantages Firstly it is not necessary to have the teacher robot learn how to follow the maze before we start we can provide it with an algorithm tried and tested in other experiments Secondly and very importantly the learner robot does not need any understanding of how the teacher perceives the world it learns what to do based on how it perceives the teachers actions the world around it and its own actions it is learning on its own terms This is a very clear indication of how combining perception and action can produce useful behaviourMoreover the independence between teacher and learner means that dierent robots can be used with dierent morphologies and dierent sensory capabilities The robot controller based on imitative learning which we implement is not entirely ethologically faithful However we expect that the insight gained from such experiments will allow us to model ethological mecha nisms in the future We do expect this type of controller to provide us with more versatile robots which learn quickly and which are relatively insensitive to small changes in the environment Moreover programming nominally identical robots will become easier It is not sucient simply to copy a program from one robot to the next no two robots are identical no two sensors the same Having one robot imitate another will obviate this problem Learning by imitation in this fashion also opens the possibilities of cross modal learning arobot which starts work when the sun rises could teach another to start work when it hears the birds singing.

Acknowledgments.

We would like to acknowledge the support of the Department of ArticialIntelligence University of Edin burgh for the provision of computing aboratory and workshop resources Ben Hope was built under SERC grant GRF Thanks are also due to John Hallam Anton Dil and Andrew Fitzgibbon for comments on drafts of this paper and technical advice.

MICHATRONICS DEPARTMENT FIELD

1.1-Features.

DC servo controlled 6-axis plus gripper
Emulates industrial robot programming techniques
Originally designed for use by the Open University
3 levels of controlling software:
level 1- posture
level 2- XYZ
level 3- linked run
Total reach of 500mm
Open construction, both hardware and software
3D simulation software available
Experiment kit (MA2000a) and project manual available to provide a structured approach to robotics
IBM XT, IBM AT software available (computer not supplied)
Two year warranty
Complies with the latest relevent EU Safety Directives

1.2-Range of Experiments.

The additional experiments kit, MA2000a, comprises gripper jaws, tools and accessories necessary to carry out the experiments outlined in the associated experiment book. The experiment book details a series of experiments to be carried out using the experiments kit. Each experiment
has a clearly defined objective and the series demonstrates the correct program techniques for various industrial applications of robots.

1.3-The experiments include:

Robot safety.

Moving the arm by lead-by-nose and by the keypad .
Continuous path programming, e.g. paint spraying .
Stacking and palletizing.
Use of sensors for component lncation .
Motor current monitoring for sorting by weight.
Simple assembly.
Drilling and de-burring .
Off-line co-ordinate programming.
Dynamic tuning.

2.1-Product Description.

The MA2000 consists of a robot arm, electronic controller interface, teach keypad and user manual. It is suitable for use with the IBM XT/AT or compatible PC (not included).
The robot is a jointed arm, with a reach of 500mm. It has three major axes: waist shoulder and elbow and three minor axes: pitch, yaw and roll. All axes are driven by d.c. motors under full Proportional, Integral and Derivative (PID) control.
The robot is also fitted with a pneumatic gripper to which a variety of jaws and tools can be attached in order to carry out specific tasks or experiments. (A compressed air supply is required to operate the gripper).The MA2000 controller interface controls the arm and is fitted with four input and output ports, for controlling and connecting with other devices. The keypad is used to communicate with the robot system. The commands on the keypad are used to control the robot's position, speed, mode of operation, program storage and retrieval

2.2-The sophisticated software allows the robot to be programmed using the most common methods found in industry.

Level 1- posture
Lead-by-nose, continuous path
drive through (teach keypad), point -to-point
lead-by-nose, point-to-point
Off line from computer keyboard
Level 2 - XYZ
as level 1 plus XYZ co-ordinate programming
Level 3 - linked run
as level 2 but with the facility to link various sequences into one continuous program

2.3-Software Features.

The software enables up to 250 program steps to be taught. During playback of a sequence of program steps, joint-space interpolation is used for movement from each point to the next.
Each step in a sequence has two associated data grids: the command data grid and the position data grid. The command data grid contains: rate of movement, mode of movement, input state expected, output state to set, wait time and step number to jump to, whilst the position data grid gives the posture value that each limb must attain and the state of the gripper (open or closed).
The rate of movement is varied in steps between 1 and 9. The maximum, 9, corresponds to a slew rate on the major axis of 45?sec and on the minor axis of 90?sec

2.4-There is a choice of 11 modes of movement at each step.

The 11 control modes are:

1.CONDITIONAL JUMP - jump to a specified step conditional to input status (MODE 1).
2.MOVE ABSOLUTE - move directly to limb positions set in position data grid or jump unconditionally to a specified step (MODE 2).
3.MOVE RELATIVE - move limbs relative to current position (MODE 3).
4.SEARCH - search a path until specified input is present, otherwise complete path and move to next step (MODE 4).
5.SEARCH AND LEARN - as mode 4 but memorise the position at which the input was received, for use in subsequent steps of the sequence (MODE 5).
6.CONTINUOUS PATH - follow a trajectory, which has been taught using lead-by-nose (MODE
7.USER FUNCTION - branch to the procedure reserved for user defined tasks and carry out the user defined BASIC program (MODES 7 and 8).
8.REPORT - either current positions, powers or protocol errors in a value derived from the analogue to digital conversion (MODE 9).
9.ADAPTIVE TUNING - change the 3 term (PID) feedback control for any specified axis (MODE 99).
10.CHAINING OF SEQUENCES - this is level 3 software and permits taught sequences of up to 250 steps each, to be linked together to run as one continuous sequence (MODE 10).
11.MULTIPLE INPUT/OUTPUT allows more than one input to be read or more than one output to be set on one sequence step. It can only be used with the controller 4 way l/O.

2.5-Supporting Teaching Material.

The robot operating manual gives detailed operating instructions and a demonstration sequence is supplied which is used as a tutorial for basic operation. The MA2000a experiments kit is available to provide a series of assignments for students

3. Specification.

3.1- Robot Mechanis

1.Configuration: A revolute arm with 3 major and 3 wrist axes.
2.Major axes: Waist, shoulder and elbow moving through 270?at maximum slew rate of 45?sec.
3.Wrist axes: Pitch, yaw and roll moving through 180?at maximum slew rate of 90?sec.
4.End effector: Pneumatically powered gripper attached to the roll axis. (A separate air supply is required).
5.End effector speed: 9 programmable speeds. Maximum speed not greater than 400mm/sec for safety considerations.
6.Reach: Nominally 500mm with jaws supplied in experimental kit (MA2000a Mk II).
7.Load capacity: 1kg dead lift at 480mm from waist axis (excludes wrist axis).
8.Drive system: Electric d.c. servo motors under closed loop, 3-term control with direct position feedback on each axis measured to 12 bit resolution.
9.Resolution: Each axis has a teach resolution of 1 part in 1000 over the angular span.
10.Repeatability: Better than +/-2mm.

11.Joint position transducers: Plastic film potentiometers with linearity of ?.25% (main axes only).
12.Sensor supply: Arm pre-wired to accept microswitch or optical sensors at the gripper.

3.2-Controller Interface.

1. Robust 19" standard housing.
2.A to D converter, 12 bit resolution. Microprocessor implementing 3 term control.
3.Manual test facility.
4.Motor drive circuits.
5.Out of limits indication.
6.I/O ports: 4 outputs, each a relay contact pair switching 1A at 24V dc.
7.4 inputs, each operating on connection to earth (ground) potential.
8.Safety: Emergency stop button, "watch dog" timer and window detector circuits.
9.Motor braking relay provides failsafe "set" of major axis movements on interruption of power supply.

3.3-Main Operating Software.

Number of steps: Up to 250 taught steps in point-to-point operation, or one block of continuous path data can be memorised.
Step commands: Position, speed, mode of movement, input and output state, wait time, branch or jump instruction. Modes of movement: (available at each step).
Eleven different modes of control are available:
1. Conditional jump
2.Move absolute
3.Move relative
4.Search
5.Search and learn
6.Continuous path
7.User function
8.Report
9.Adaptive tuning
10.Chaining of sequences
11.Multiple input/output - (allows more than one input to be read or more than one output to be set on the controller 4 way l/O or the expansion l/O interfaces).
Teach Keypad.
20 keys: membrane type with function legends.

3.4 Teaching Methods.

Drive through, using the teach keypad, for point-to-point operation.
Lead-by-nose, for continuous path or point-to-point operation.
Off line by computer: XYZ and joint position.

4.1 Ancillaries.

The following products are available for use with the MA2000 robot system.MA2000a Mkll Experiments Kit. MA2000x Software to allow taught sequences (up to 250 steps each) to be chained together and to allow other programs to share the same host computer. MA9080 "Workspace" 3D robot simulation software for IBM or compatible computer running Windows '95.
Services Required
Single phase electrical supply of 240 V, 50 Hz; or 110 V, 60 Hz is required. Please specify voltage frequency and tolerance on ordering. A clean dry air supply of approximately 3 bar is required to operate the gripper. This may be provided by an air bottle or a suitable compressor.
Noise
The measured sound pressure level of this apparatus is less than 70 dB(A)
Dimensions &Weights
Gross:0.3 m; 50 kg(approx - packed for export)

4.2-Tender Specification .

A dc. servo motor robot with 6 axes of movement and a pneumatic gripper. The system to consist of a robot arm, electronic controller interface with 4 input and 4 output ports.

The arm to have a reach of 500mm and a dead lift capability of 1kg at 480mm from the waist axis (wrist excluded).
The operating software to enable the robot arm to be programmed using: lead-by-nose, continuous path, drive through point-to-point, lead-by-nose point-to-point and off line programming, in both robot joint space and real world XYZ co-ordinates.
Optional 3D simulation software with solid modelling. Supplied with a two year parts and labour warranty. Requires an IBM PC XT or better (not included).

Divorce - How to rebuild your life, color your hair and move on

Did you know that when you go through a life changing experience you are likely to go to the beauty salon before you do anything? Most women will actually seek a different hair style when something major happens in their life. You will want to quickly seek comfort for your divorce and then you have to stand on your own. You may want to go to the hair salon once you have made the announcement of the split and have faced the situation. Once you have recognized that you are going through a divorce you will need all the people who love you around so that you are able to get all the support that you need to help rebuild your life and move on.
The first step is to know who you can count on and who can't. This is when your real friends became noticeable. You need to have people who care to support your grieving and to help you find ways to move on. Most of your so-called friend will say everything will be fun, don't worry. When someone allows your feelings to be written off like that, they are not your friends at all. You will want to make sure that you understand the difference between real friends and people who claim to be your friend.
Once you have found the support and strength to move on, you will feel the need to make some drastic changes. You may want to change the color of your hair or you may just want to change your wardrobe or encourage a change in behavior. You should try things that you never would have gotten to do with your soon to be ex. You should also think about making changes about your appearance that they wouldn't necessarily encourage from you. You will want to do things like change your hair color, change your style of hair, or get an extreme haircut.
If you have never had short hair, you may want to try it. When you cut or when you dye your hair you will find empowerment. You will feel like you have control over yourself and your life. The truth of the matter is that you do have all the control. You have the right to change and you have the right to do whatever it is that you want or find that makes you happy. Before you allow yourself to fall in depression, you may want to start thinking about what has happen, what has changes, and what you would like to do as a result of.
Obviously, there were many factors that made you and your partner to split, but you don't have to be someone that you aren't. Over the years or course of your marriage, you probably gave up a lot. You most likely changed because they encourages you to become exactly what they wanted, but you are no longer in that relationship and you can begin to do the things that you love once again.
Any woman that can go through a divorce and survive it is a very strong person. Some women will collapse and fall into a deep depression. Go out and do everything for yourself. If you have always wanted to be a blonde, give it a shot. If you have ever wanted to cut your hair, go for it. Once you begin to do things that you normally wouldn't do you will find liberation. You will have liberation from all the chains of marriage. You will have the freedom to be yourself and show your new freedom in anyway that you would like.

History of Some Christmas Traditions Christians celebrate Christmas to observe the birth of Jesus Christ, which is an event and not a tradition. But many other activities related to celebrating the Christmas season evolved from certain traditions, many of which are from other countries, particularly from peoples in Europe. Among common items used in Christmas decorations are the holly and the mistletoe. Both are used primarily in wreaths and garlands. The Druids started the tradition of using the mistletoe as decorative items up to two hundred years before Christ. To celebrate the winter season, the Druids would gather the plants and use them to decorate their homes. The Druids believed the mistletoe would bring good luck and ward off evil spirits. They also believed that the mistletoe had a healing quality and could be used for everything from healing wounds to increase fertility. In Scandinavia, the mistletoe was seen as a plant of peace and harmony and was associated with Frigga, the goddess of love. This association is probably what led to the custom of kissing under the mistletoe. In the Victorian period, the English also would hang mistletoe from ceilings and in doorways during holidays. The habit developed that if someone was standing under the mistletoe, someone else in the room would kiss that person. Such outright behavior was not generally seen in Victorian society. The use of the mistletoe in Christmas celebrations was once banned by the church however because of its associations with pagan traditions, and the use of holly was suggested as a substitute. Poinsettias are another traditional decorative flower used at Christmas. It is native to Mexico and is named after Joel Poinsett, who was the first U.S ambassador to Mexico and who brought the plants to America in 1828. Mexicans believe the plants were a symbol of the Star of Bethlehem and that's one reason they are associated with Christmas. There's also the story that a young boy was going to see the Nativity Play at a church but realized he didn't have a gift for Baby Jesus. The boy gathered some green branches, which others scoffed at. But as he placed them near the manger, a bright red poinsettia flower started to bloom on each branch, which gave rise to their traditional use at Christmas. Candy canes became a Christmas tradition not because their red and white stripes matched the colors of the season, but for the most unusual reason of discipline. that's because they were first used as treats that were give to German children to keep them well-behaved for the duration of church sermons. Over time, the legend of candy canes at Christmas came to be associated with some of the strongest symbols and beliefs of Christianity: the Father, Son and Holy Ghost known as the Trinity, the Blood of the Son of God, Jesus as the embodiment of holiness, purity and without sin and the Son of God as the shepherd of man. The candy cane represents these symbols respectively with its three stripes, its red and white color and its shape. Sending greeting cards during Christmas and the holidays is as prevalent today as the custom of giving gifts. The tradition of sending Christmas cards started in 1840 in Britain with the start of public postal delivery service of the 'Penny Post.' Then from about 1860, large numbers of Christmas greeting cards started to be produced. The popularity of the cards increased in Britain when they could be sent by the postal service for one half-penny, which was half the price to post a standard letter at the time, if they were in an unsealed envelope. Religious pictures of Mary, Joseph , Baby Jesus, the angels, shepherds and Wise Men were traditionally placed on Christmas cards. Some cards today include scenes from the Nativity, but pictures of Santa Claus, winter scenery, Christmas trees, gift packages and others are also depicted on contemporary Christmas greeting cards.

Choosing Batting Quilt Fibers for Craft How to choose batting fibers

Quilts include the crib sizes, twin, and full, double fit, queen, and king. The standard crib fit is around 45-inches time’s 60. Twin fits are 72 x 90, full and the double are 81 x 96, queen fits are 90 x 108, and the king fits are around 120 x 120.
To choose your materials you must consider batting quilt fibers. Once you choose your batting make sure that, you unfold the cotton material and let it set a couple of days before you start crafting. The batting will relax and inflate. You want to space your batting closely to avoid bunching also when crafting your quilt. Some materials require pre-washes before you can use the fabric. Read your labels.
You have options in battings, including the traditional, which is often made of cotton and the polyesters. The blends of polyester and cotton will shrink sometimes. To stitch the cotton you will need to create intervals of larger stitches, yet if you combine polyester with your cotton, you can minimize the stitches.
The line of battings, include polyester, silk, wool, etc as well. If you choose the polyester, you can create a non-shrinking quilt with intervals of larger stitches. In addition, you can create intervals of wider expansions, which you can craft your quick at a speedier pace. Polyester is the choice of battings, since crafters can design a quick, machine washable, and non-shrinking quilt. As well, the crafter can design a thinner quilt verses the thicker, since polyester is a batting made of “high loft.”
My favorite is silk, yet if you are creating a traditional style quilt, the silk may not be suitable. You can still make a quilt of silk, yet you will pay top-dollar and spend a length of undesired time to finish your project. In fact, most crafters do not recommend silk for creating quilts.
Wool has migrating fibers, yet you can sew through the material with ease. You will need to space closely when needling. The wool over time will loose its fibers however. Wool will also fuzz. You can use lightweight materials, or cloths to prevent fuzziness, as well as to prevent fiber loss. Wool is not suitable for machine wash; rather you should take your quilt to a professional cleaner, or wash it by hand and allow it to air dry.
Now choose your style: If you want the antique or traditional quilt, you will need to use the “low loft” material. The quilts include the Fairfield, which is 100% bleached cotton, the poly-filled cottons, which is 80 % cotton; the “Mountain Mist/Blue Ribbon Stearns” are 100% cotton as well. You can also choose the 100% polyester, Morning Glory, or the Glory BEE I, which is also 100% polyester. Many other styles and varieties are available.
Once you decide which fibers, or fabrics you want to craft your quilt you can then consider your backing. You can purchase yards of backing. The backing today makes it easy to fill in the length and width of your quilt. Ultimately, if you choose backings that come up short or longer you can stitch a couple of pieces together to even your quilt. Still, you want to make sure that the backings work in harmony with your fabrics, or fibers.
Ultimately, if you want to spend time making a quilt you can choose blocks and patches. The pieces of material were frequently used by grandmas, or women of traditional days. The quilts are often sturdier than the modern quilts, yet you can still craft a strong quilt today.

Craft Stitching Porcelain Doll Sleeves How to stitch porcelain doll sleeves
Once you have begun making your dress, you want to stitch your porcelain doll sleeves. To get started, affix the lace, meeting it with the edges of the sleeves and crisscross. Press once you finish. Next, gather the dual rows of your stitch and continue about the crown of the sleeve until it fits into the right armhole, coming together, pull the collected fit up, and stitch them collectively whilst keeping your face liberated. Do the same to complete the opposite sleeve.
Starting at the right sleeve joint with the bodice, sew the seams of the underarm from the edges of your sleeve and from side to side seams of the upper region of the dress. Now begin stitching the seams at the side of your bodice so that it faces jointly and moves to face a different direction within, covering the side facing seams. You may need to cut to fit the areas around the seams of the armholes, including the facing holes. Change directions, turning in the hems about the facing armhole, match the shoulders as well as the seams at the side, and then “slip” suture the facing in the region of the armholes, moving in the opposite direction as you stitch. Use the elastic hat and fasten it to the interior region to create the starting legs of your underclothing. You may need to cut to fit, yet add glue before you begin cutting.
Now you have completed your sleeves for your porcelain doll. Once you finish your sleeves, you may want to design and elegant skirt to fit your doll, as well as an apron. To get started with your skirt finish your patterns at the untreated edges, and at the seams of the back using the crisscross stitching method. Next, sew the seams at the back from the dot and to the hems. Line up dual lines and gather your stitches about the crown of your skirt. Fold the back seams and permit to the left side on mutual sides of your seams at the back.
The bodice and front middle of your skirt should come together, as well as the fold lines at the back of the upper region of your dress. Extend to the opening at the back of your skirt and keep the facing bodice liberated. Collect your thread by pulling up and extend to fit the skirt connecting it to the bodice and distributing the collected sections uniformly. Next, trim or shape the seams and fold an upward hem on the facing upper region of your dress so that it corresponds with the seams at the side of your bodice. Use the “slip stitch” method and stitch the seams along the facing so that it connects with the bodice and the skirt.
Now you are ready to dress your doll. As you put the dress on the doll note any areas that may need length added, and mark the seam lines. Finish the dress at the untreated edges of your hem and crisscross. Next, turn the width to needed size and hem while using the slip suture method to fit the skirt. You can make buttonholes next. To start hand sow or machine stitch your buttons after adding glue to the fabric to hold it together. Use a pin to make your buttonholes. Allow the glue to dry and then cut the region, using craft scissors. The buttons or press “000 studs” can be used and sown at the back of your skirt.
You are now ready to create an apron to fit your porcelain doll dress.