This report discusses robot reliability which includes the basics of robotics, history of robots, basic components of robot, robotics applications, and classification of robots. Furthermore, it includes detailed discussion on robot reliability taking into account the reliability concepts and mathematics, failure causes and classifications, robot reliability measures and robot reliability analysis methods. After that, it discusses the important subject of robot maintenance in context of robot reliability as poor maintenance leads to poor reliability. Final chapter discusses the failure data and analysis that includes the reliability related data sources, and the hazard plotting method which is important and vital for robot reliability studies.
Robots are increasingly being used in the various types of industry carrying out different types of tasks which gives weight to the fact that the humans have become dependent on robotics for carrying out work efficiently thereby saving time and effort. For that to happen, a robot has to be reliable and safe otherwise it may lead to inconvenience and high maintenance costs in addition to safety hazards they might pose. As robots are made up of electrical, mechanical and hydraulic parts, making robots reliable and safe is a daunting and a challenging task due to many causes of failures. In order to make significant improvements in robot reliability, we need to have a qualitative as well as a quantitative understanding of the methods which can be used for measuring and discussing reliability and failure of robots, that would help us in providing solutions that can help in tackling issues related to robotics.
Table of Contents
According to Robot Institute of America "A robot is a reprogrammable, multifunctional manipulator designed to move material, parts, tools or specialized devices through variable programmed motions for the performance of a variety of tasks"
According to the International Organization for Standardization under ISO 8373, a robot is: "An automatically controlled, reprogrammable, multi-purpose manipulator programmable in three or more axes, which may be either fixed in place or mobile for use in industrial automation applications."
According to the Oxford English dictionary "Robot is machine capable of carrying out a complex series of actions automatically, especially one programmable by a computer"
According to the free dictionary "Robot is a mechanical device that sometimes resembles a human and is capable of performing a variety of often complex human tasks on command or by being programmed in advance" or "A machine or device that operates automatically or by remote control" or "A person who works mechanically without original thought, especially one who responds automatically to the commands of others"
1.2 History of Robots
Ancient history concerning robots can be traced back to almost 5000 years to when Egyptians built water powered clocks and the Chinese and Greek built water and steam powered toys. However, Greeks were the first one who came with the idea of a functional robot which can be seen in the writings of Aristotle in the fourth century B.C. in which he wrote "If every instrument could accomplish its own work, obeying or anticipating the will of othersâ€¦" which took nearly 2000 years to put the idea into practice.
In the modern era, the first programmable looms controlled by punch cards were developed in France in the 1720's. In 1770, Swiss tradesmen human-like automatons that could write, draw, and play musical instruments. In 1818, Mark Shelly published the first true science-fiction novel called "Frankenstein" which was about a machine monster created by humans.
The word "Robot" meaning "worker" was first used in a play written by Karl Capek (1890-1938), a Czechoslovak science-fiction writer, which opened in London in 1921. Isaac Asimov then wrote a series of stories about robots at the age of 19 in 1939 , nine of which were published in a book called " I Robot" In 1950. In 1942, Asimov developed the following "three laws of robotics" :
A robot may not injure a person nor, through inaction, allow a person to come to harm
A robot must always obey orders from people except in circumstances in which such orders are in conflict with the first law
A robot must protect its own existence except in circumstances in which it is a conflict with the above two laws.
In 1946, George Devol developed magnetic process control which was a general purpose analogue storage device. The first commercially available robot was sold by Planet Corporation in 1959. A year later in 1960, the Unimate robot was developed from Devon's original device. By mid 1960's, robotic research laboratories were established at Stanford Research Institute, MIT and at the University of Edinburgh. In 1967, Japan, today a leading user of robots, imported its first robot. Currently, the worldwide industrial robot population is estimated to be around one million.
1.3 Classification of Robots:
The Japanese Industrial robot association (JIRA) classifies robots into six categories:
Manual handling device
Fixed sequence robot
Numerical control robot
Variable sequence robot
1. Intelligent robot
Intelligent robot has the capability of understanding its local environment and successfully accomplishing a task despite variations in the surrounding conditions.
2. Manual handling devices
Manual handling devices are different in that they have the freedom of being actuated by human operators.
3. Playback robots
Human operators also play an important role in playback robots, by performing the task in question manually by leading or controlling the robot recording the trajectories. At the moment of need the recorded information is recalled enabling the robot to carry out the same task in its automatic mode.
4. Fixed Sequence Robots
Robots belonginf to this category perform the successive stages of a task according to a predetermined and rigid approach which is difficult to change.
5. Numerically Controlled Robots
In numerically controlled robots, the operator furnishes robots with a movement program instead of teaching them the task through manual means.
6. Variable Sequence Robots
Robots belonging to this category are basically the same as the fixed sequence robots, difference being that the successive stages can be changed with ease.
1.4 Production Robots
These robots may be classified into two broad classifications:
Work item-handling robots
a). Work item-handling robots:
The work item-handling robots are composed of appropriate grippers for the work items to be handled. Generally, the task performed by the work piece handling robots is to fetch work items from a specified initial position to a specified final position.
Examples of the areas in which this type of robot perform tasks are :
Investment casting and
b). Tool-handling robots:
The tool-handling robots make use of special tools instead of grippers or safely grapple such tools with their grippers.
Measuring instruments for inspection
Finishing paint sprayers
1.5 Robot Structure Categories
A robot structure may be classified into four categories as shown in the fig below:
Categories of robot structure
The robot arm has one primary function, to translate or move both the wrist and the end-effector to a certain point in the surrounding work zone.
The primary responsibility of the wrist is to provide the desired approach and orientation. There are many types of wrists available on the market, most of which have three degrees of freedom: role, pitch and yaw.
The body of a robot is the frame on which its arm is installed.
Generally, an industrial robot is composed of the following:
a stationary body
a wrist, and
In order to allow additional degrees of freedom, a moving body part may be included.
The end-effector is the terminal part of an industrial robot, and is composed of a hand like device having multiple fingers, vacuum cups, or a tool similar in task in question. Functional characteristics of an end-effector are:
To grapple and manipulate tools and work items; and
To develop interactions between work items by welding, assembly, spraying
1.6 Timeline history of classification of Robots
1.6.1 Research or Service Robots
History of service robots is as follows:
1946: George Devol patents a general purpose playback device for controlling machines. It uses a magnetic process recorder. American scientists J. Presper Eckert and John Mauchly build the first large electronic computer called the Eniac at the University Pennsylvania. The second computer, the Whirlwind, solves a problem at M.I.T. The Whirlwind is the first general-purpose digital computer.
1948: Norbert Wiener, a professor at M.I.T., publishes his book, Cybernetics, which describes the concept of communications and control in electronic, mechanical, and biological systems.
1950: Alan Turing publishes Computing Machinery and Intelligence in which he proposes a test to determine whether or not a machine has gained the power to think for itself. It becomes known as the "Turing Test".
1951: A teleoperator-equipped articulated arm is designed by Raymond Goertz for the Atomic Energy Commission.
1959: John McCarthy and Marvin Minsky start the Artificial Intelligence Laboratory at the Massachusetts Institute of Technology (MIT). Computer-assisted manufacturing was demonstrated at the Servomechanisms Lab at MIT. Planet Corporation markets the first commercially available robot.
1960's: Johns Hopkins creates the beast. It is controlled by hundreds of transistors and able to seek out photocell outlets when its battery runs low.
1960: The General Electric Walking Truck was a 3,000 pound, four-legged robot that could walk four miles an hour. It was powered by a computer. Ralph Moser developed the machine.
1961: Heinrich Ernst develops the MH-1, a computer operated mechanical hand at MIT.
1963: John McCarthy leaves MIT to start the Artificial Intelligence Laboratory at Stanford University. The first artificial robotic arm, the Rancho Arm, was controlled by a computer and designed as a tool for the handicapped. Its six joints gave it the flexibility of a human arm.
1964: Artificial intelligence research laboratories are opened at M.I.T., Stanford Research Institute (SRI), Stanford University, and the University of Edinburgh.
1966: The Stanford Research Institute (later to be known as SRI Technology) creates Shakey the first mobile robot to know and react to its own actions.
1968: The octopus-like Tentacle Arm was developed by Marvin Minsky.
1969: The Stanford Arm was the first electrically powered, computer-controlled robot arm. Designed by Victor Scheinman, a Mechanical Engineering student, this arm's design becomes a standard and is still influencing the design of robot arms today.
1970's: Scientists at Edinburgh University create the Freddy robot, taking steps in hand-eye coordination technology. This first assembly robot constructed a toy boat and car from a heap of mixed parts tipped onto a table.
1976: Robot arms are used on Viking 1 and 2 space probes. Vicarm Inc. incorporates a microcomputer into the Vicarm design.
1979: The Standford Cart crosses a chair-filled room without human assistance. The cart is equipped with a television camera mounted on a rail that takes pictures and relays them to a computer so that distances can be analyzed.
1983: The Remote Reconnaissance Vehicle became the first vehicle to enter the basement of Three Mile Island after a meltdown in March 1979. This vehicle worked four years to survey and clean up the flooded basement.
1984: The CoreSampler drilled core samples from the walls of the Three Mile Island basement to determine the depth and severity of radioactive material that soaked into the concrete.
1984: The Terregator pioneered exploration, road following and mine mapping. It was the world's first rugged, capable, autonomous outdoor navigation robot.
1985: REX was the world's first autonomous digging machine. It sensed and planned to excavate without damaging buried gas pipes. This robot used a hypersonic air knife to erode soil around pipes.
1986: Honda begins a robot research program that starts with the premise that the robot "should coexist and cooperate with human beings, by doing what a person cannot do and by cultivating a new dimension in mobility to ultimately benefit society."
1988: The Pipe Mapping computes magnetic and radar data over a dense grid to infer the depth and location of buried pipes. This outperforms hand-held pipe detectors.
1989: A walking robot named Genghis is unveiled by the Mobile Robots Group at MIT. It becomes known for the way it walks, popularly referred to as the "Genghis gait".
1988: The Locomotion featured a chassis that steers and propels all wheels so that it can spin, drive, or spin while driving. Its software can emulate a tank, car or any other wheeled machine.
1990: The Ambler was a walking robot that enables energy-efficient overlapping gaits. Developed as a test bed for research in walking robots operating in rugged terrain.
1992: Dante I rappels mountain sides using a spherical laser scanner and foot sensors. It entered the crater of Antarctica's Mt. Erebus but did not reach the lava lake.
1992: NavLab II was the automated HUMMER that pioneered trinocular vision, WARP computing, and sensor fusion to navigate off-road terrain.
1993: Honda debuts the P3, the fruit of its decade long effort to build a humanoid robot.
1994: The Dante II, build by CMU Robotics, samples volcanic gases from the Mt. Spurr volcano in Alaska.
1997: NASA's PathFinder lands on Mars and the Sojourner rover robot captures images.
2003: NASA launches the MER-A "Spirit" rover destined for Mars and the MER-B "Opportunity".
1.6.2 Time line History of Industrial Robots
History of Industrial robots is as follows:
1954: The first programmable robot is designed by George Devol. He coins the term Universal Automation.
1956: Devol and engineer Joseph Engelberger form the world's first robot company, Unimation.
1960: Unimation is purchased by Condec Corporation and development of Unimate Robot Systems begins. American Machine and Foundry, later known as AMF Corporation, markets a robot, called the Versatran, designed by Harry Johnson and Veljko Milenkovic.
1962: The first industrial robot was online in a General Motors automobile factory in New Jersey. It was Devol and Engelberger's UNIMATE. It performed spot welding and extracted die castings.
1969: Nachi starts its robotic business
1973: German robotics company, KUKA, creates the first industrial robot with six electromechanically-driven axes. It is called the Famulus.
1974: A robotic arm (the Silver Arm) that performed small-parts assembly using feedback from touch and pressure sensors was designed. Professor Scheinman, the developer of the Stanford Arm, forms Vicarm Inc. to market a version of the arm for industrial applications. The new arm is controlled by a minicomputer.
1974: Industrial robots were developed and installed in Fanuc factory. Dr. Inaba, President of FANUC was rewarded with "the 6th Annual Memorial Award of Joseph Marie Jacquard" by the American NC Society. The production and sale of DC servo motors were started under GETTYS MANUFACTURING CO., INC license.
1977: The Motoman L10 was introduced. It featured five axes and a maximum workload of 10 kg, which included the gripper. It weighed 470kg. The Motoman L10 was the first robot that Yaskawa introduced on the market.
1977: ASEA, a European robot company, offers two sizes of electric powered industrial robots. Both robots use a microcomputer controller for programming and operation. Unimation purchases Vicarm Inc. during this year.
1978: Vicarm, Unimation creates the PUMA (Programmable Universal Machine for Assembly) robot with support from General Motors. Many research labs still use this assembly robot.
1979: Nachi developed the first motor-driven robots for spot welding.
1979: OTC DAIHEN was known as OTC America. OTC was an acronym for the Osaka Transformer Company. Located in Charlotte, NC, OTC was originally a supplier of welding equipment for other transplant companies. They expanded to become a provider to the Japanese auto market of GMAW supplies. In these early years, OTC Japan introduced its first generation of dedicated arc welding robots.
1980: The industrial robot industry starts its rapid growth, with a new robot or company entering the market every month.
1981: Takeo Kanade builds the direct drive arm. It is the first to have motors installed directly into the joints of the arm. This change makes it faster and much more accurate than previous robotic arms.
1985: OTC DAIHEN became the official OEM supplier of robots to the Miller Electric Company. Miller chose to assign different model numbers to the robots sold in the North American market. The prefixed the letters in the model with "MR," for Miller Robot. Miller no longer supports the robots that were manufactured in this era. The Japanese models featured their own number and name.
1987: ASEA of Vasteras, Sweden (founded 1883) and BBC Brown Boveri Ltd of Baden, Switzerland, (founded 1891) announce plans to form ABB Asea Brown Boveri Ltd., headquartered in Zurich, Switzerland. Each parent will hold 50 percent of the new company.
1988: The Motoman ERC control system was introduced with the ability to control up to 12 axes, more than any other controller at the time.
1989: Nachi Technology Inc., U.S.A. is established.
1992: FANUC Robot School was established. GM Fanuc Robotics Corporation was restructured to FANUC's wholly owned share holding company, FANUC Robotics Corporation, together with its subsidiaries, FANUC Robotics North America, Inc. and FANUC Robotics Europe GmbH. A Prototype of the intelligent robot was built.
1994: The Motoman MRC control system was introduced with the ability to control up to 21 axes. It could also synchronize the motions of two robots.
1995: Miller departed from the robotic business. OTC launched the Dynamic Robotic Division and moved the headquarters to Ohio to focus on selling robots to new users.
1996: Nachi expands robotics business, cutting tool, and bearing product ranges.
1998: The introduction of the XRC controller allowed the control of up to 27 axes and the synchronized control of three to four robots. The Motoman UP series introduced a simpler robot arm that was more readily accessible for maintenance and repair. Honda was instrumental in driving the development of both the UP series of arms and the XRC arm control.
2003: OTC DAIHEN introduced the Almega AX series, a line of arc welding and handling robots. The AX series robots integrate seamlessly with the OTC D series welding power supplies for advanced control capabilities.
1.6.3 Timeline History of Educational Robots
History of Educational robots is as follows:
1977: Star Wars is released. George Lucas' movie about a universe governed by the force introduces watchers to R2-D2 and C-3PO. The movie creates the strongest image of a human future with robots since the 1960's and inspires a generation of researchers.
1986: LEGO and the MIT Media Lab collaborate to bring the first LEGO based educational products to market. LEGOs are used by in the classrooms of thousands of elementary school teachers.
1986: The Society of Manufacturing Engineers Robotic Technology and Engineering Challenge (now known as the National Robotics Challenge) was started under the guidance and inspiration of Tom Meravi, Associate Professor from Northern Michigan University and the late Dr. James Hannemann.
1989: FIRST (For Inspiration and Recognition of Science and Technology) was founded by Dean Kamen to inspire young people's interest and participation in science and technology.
1992: In an attempt to build a radio controlled vacuum cleaner Marc Thorpe has the idea to start a robot combat event.
1992: FIRST holds its first competition in a New Hampshire high-school gym. There are 28 teams to compete in the "Maize Craze" challenge.
1994: Marc Thorpe starts Robot Wars at Fort Mason center in San Francisco, CA.
1995: The second annual Robot Wars event is held at Fort Mason Center, San Francisco, CA.
1998: Tiger Electronics introduces the Furby into the Christmas toy market. It quickly becomes "the toy" to get for the season. Using a variety of sensors this "animatronic pet" can react to its environment and communicate using over 800 phrases in English and their own language "Furbish".
1998: LEGO releases their first Robotics Invention System 1.0. LEGO names the product line MINDSTORMS after Seymour Papert's seminal work of 1980
1999: LEGO releases The Robotics Discovery Set, Droid Developer Kit and the Robotics Invention System 1.5.
2000: Humanoid robots, Honda Asimo, Sony Dream Robots (SDR), and the AIBO robot dog are showcased.
2000: The Battlebots event is held in Las Vegas, Nevada.
2000: LEGO releases the MINDSTORMS Robotics Invention System 2.0
2001: LEGO releases the MINDSTORMS Ultimate Builder's Set
2001: The SME announced the end of sponsorship for the Society of Manufacturing Engineers Robotic Technology and Engineering Challenge.
2002: Honda's ASIMO robot rings the opening bell at the New York Stock Exchange.
2003: SONY releases the AIBO ERS-7, its 3rd generation robotic pet.
2004: The humanoid, Robosapien is created by US robotics physicist and BEAM expert, Dr. Mark W Tilden.
2004: The previously held Society of Manufacturing Engineers Robotic Technology and Engineering Challenge evolves into the National Robotics Challenge under the leadership of Marion, OH, educators Ed Goodwin, Ritch Ramey, and Tad Douce.
Robot reliability is very complex subject where there are numerous interlocking variables in evaluating and accomplishing various reliability levels. A successful robot installation has to be safe and reliable. A robot with poor reliability leads to a number of problems such as high maintenance costs, unsafe conditions, inconvenience etc.
To get rid of such problems, the American National Standard for Industrial Robots and Robot Systems- Safety requirements clearly highlights and calls for the design and construction of robots in such a way that any single, foreseeable failure will not lead to robots hazardous motion. There are many different parts which are used in robots i.e., electrical, electronics, hydraulic, mechanical which in turn makes the task of producing highly reliable robots a challenging one. Furthermore, the environments in which the robots have to operate may not be suitable and may vary between installations for same or identical models. 
The expected useful life of robots is around 40,000 operating hours and the mean time between failures (MTBF) at least 400 hours. In addition, the mean time to repair (MTTR) of 8 hours or less is desirable. The best MTBF achieved for robots so far is only of the order of 2500 hours. The yearly cost of maintenance associated with robots is approximately 11% of the procurement cost. The mechanism of the robots, instead of the electronics, accounts for most of the downtime and maintenance costs associated with robots. 
2.2 Background of Reliability
Reliability is the probability that a system will be operating properly at a given time. Another way of saying this is that reliability is the probability that no failures will occur before a given time. When evaluating the reliability of a system, we must first identify the ways in which the system may fail and then determine the probabilities of those failures occurring. Many failures are the result of human error. These errors can be classified into three categories: errors of omission, errors of commission, and operational errors 
2.2.1 Errors of omission:
These errors are related to design; i.e., the design specifications fail to define the device in a way that will allow it to function correctly.
2.2.2 Errors of commission:
These errors are related to implementation; i.e., the product fails to match the design specifications.
2.2.3 Operational errors:
These errors are related to improper use of the system, for instance, using it in an environment for which it was not designed.
What all of these errors have in common is that they can be eliminated, or at least dramatically reduced, through process improvement. Methods such as ISO 9000 Quality Management can be applied to reduce design and implementation errors, and knowledge from human factors and other disciplines can be applied to reduce operational errors. For many devices, there are additional failures that are due to inherent properties of the materials and processes used by the device. For example, the grease in a bearing will eventually dry out and the bearing will fail due to lack of lubrication. No amount of process improvement can eliminate this physical reality. What can be done is to design the bearing so that there is a very small chance of failure during the expected lifetime of the product. Alternatively, the product can be designed so that the bearing can be easily re-lubricated or replaced, and then a maintenance or replacement interval can be specified as part of the design. These alternatives apparently have different costs and most likely different probabilities of bearing failure. It is these tradeoffs among design options, probability of failure, and cost that are typically considered in the domain of reliability engineering 
2.3 Robot Failure Causes and Classifications
There are many causes of robot failures. Some of the most common ones are as follows :
Oil pressure valve problems
Printed circuit board problems
Servo valve problems
Furthermore, robot problems or troubles followed the following order:
Control system problems
Incompatibility of jigs and other tools
Robot body-related problems
Programming and operation errors
Welding gun troubles and difficulties with other tooling parts
Deterioration, precision deficiency
There are basically four types of failures (Fig 2), that affect robot reliability and its safe operation. These are random component failures, systematic hardware faults, human errors, and software failures.
Random component failures
Systematic hardware faults
Fig 2. Types of failures that affect robot reliability
Failures that occur during the useful life of a robot are called random component failures because they occur unpredictably. Some of the reasons for the occurrence of such failures are undetectable defects, low safety factors, unexplainable causes, and unavoidable failures..
Systematic hardware faults are those failures that occur because of the existence of unrevealed mechanisms in the robot design. Reasons such as peculiar wrist orientations and unusual joint-to-straight-line mode transition can lead the robot not to perform a specific task or to execute certain portions of a program.
Human errors are caused by people who design, manufacture, test, operate, and maintain robots. Various studies reveal that the human error is a significant element of total equipment failures. Some of the important reasons for the occurrence of human errors are poor equipment design, poorly trained operation and maintenance personnel, task complexity, inadequate lighting in the work areas, improper tools used by maintenance personnel, and poorly written maintenance and operating procedures.
Software failures are an important element in the malfunctioning of robots, and they occur in robots due to reasons such as embedded software or the controlling software and application software. Some of the methods that can be useful to reduce the occurrence of software faults in robots are failure mode and effects analysis (FMEA), fault tree analysis (FTA), and testing.
2.4 Robot Reliability Measures
2.4.1 Mean Time to Robot Failure
Mean time to robot failure can be obtained by using either of the following three formulas:
MTRF is the mean time to robot failure
Rrb(t) is the robot reliability at time t.
's' is the Laplace transform variable.
Rrb (s) is the Laplace transform of the robot reliability function.
TNRF is the total number of robot failures.
RPH is the robot production hours.
DDTRF is the downtime due to robot failure expressed in hours.
2.4.2 Mean Time to Robot-related Problems
Mean time to robot-related problems is the average productive robot time prior to the occurrence of a robot-related problem and is defined by:
Where, MTRP is the mean time to robot-related problems. DDTRP is the downtime due to robot-related problems expressed in hours. TNRP is the total number of robot-related problems.
2.4.3 Robot Reliability
Robot reliability is given as:
Where, Rrb (t) is the robot hazard rate or time-dependent failure rate and gives the general expression for obtaining robot reliability. More specifically, it can be used to obtain a robot's reliability when robot times to failure follow any statistical distribution (e. g., Weibull, normal, gamma, or exponential).
2.4.4 Robot Hazard Rate
Robot hazard rate is given as:
Where, Rrb (t) is the robot hazard rate and Rrb (t) is the robot reliability at time 't'
Equation can be used to obtain robot hazard rate when robot times to failure follow any time-continuous distribution (e. g., exponential, Rayleigh, Weibull, etc.)
2.5 Robot Reliability Analysis Methods
There are many methods used to perform various types of reliability analysis in the field of reliability engineering. Some of them can be used quite effectively to conduct robot reliability-related studies. Four of these methods are shown inFigure3 .
Fig 3. Methods for performing robot reliability-related studies
parts count method,
failure modes and effect analysis (FMEA) ,
fault tree analysis, and
Parts count method:
The parts count method is used during bid proposal and early design phases for estimating equipment failure rate. The method requires information on items such as equipment/product use environment, generic part types and quantities, and part quality levels.
Failure Modes and Effect Analysis:
Failure modes and effect analysis (FMEA) is an effective tool to conduct analysis of each failure mode in the system/equipment to determine the effects of such failure modes on the total system/equipment. This method was developed by the United States Department of Defense in the early 1950s and comprises of the following six steps]:
Define system/equipment and its associated requirements.
Develop appropriate ground rules.
Describe the system/equipment and all its related functional blocks.
Highlight all possible failure modes and their effects.
Develop critical items list.
Document the analysis.
Fault tree Analysis:
Fault tree analysis is a widely used method to evaluate reliability of engineering systems during their design and development phase. A fault tree may be described as a logical representation of the relationship of basic or primary events that result in a specified undesirable event called the "top event". This method was developed in the early 1960s at Bell Telephone .
The Markov method can be used in more cases than any other reliability evaluation method and is used to model systems with constant failure and repair rates.
2.6 Models for Performing Robot Reliability and Maintenance Studies
There are many mathematical models that can be used to perform various types of robot reliability and maintenance studies. Four of these models are presented below .
2.6.1 Reliability and Availability Analysis of a Robot System Failing with Human Error
This model represents a robot system that can fail either due to a human error or other failures (e. g., hardware and software). The failed robot system is repaired to its operating state. The robot system state diagram is shown in Figure 4. The numerals in the rectangle, circle, and diamond denote system states. The following assumptions are associated with this robot system model :
Human error and other failures are statistically independent.
Human error and other failure rates are constant.
The failed robot system repair rates are constant.
The repaired robot system is as good as new.
Fig 4. Robot system state space diagram
The following symbols are associated with the diagram in Figure 4 and its associated equations:
Pi (t) is the probability that the robot system is in state i at time t; for i=0 , (working normally), i = 1 (failed due to a human error), i = 2 (failed due to failures other than human errors). Î»h is the robot system human error rate. Î±h is the robot system repair rate from failed state 1. Î» is the robot system non-human error failure rate. Î± is the robot system repair rate from failed state 2.
Using the Markov method , robot system reliability is given as :
Furthermore, Mean time to Robot Failure and robot system hazard rate are given as :
The right-hand side of above equation is independent of time, which means that the robot system failure rate is constant.
2.6.2 Model II
This model is concerned with determining the economic life of a robot, more specifically, the time limit beyond which it is not economical to carry out repairs. Thus, the economic life, Te, of the robot is expressed as:
Where, Crin is the annual increase in robot repair cost. SVr is the robot scrap value and Cir is the robot initial cost (installed).
2.6.3 Model III
This model can be used to calculate the optimum number of inspections per robot facility per unit time. This information is useful to decision makers because inspections are often disruptive; however, such inspections usually reduce the robot downtime because they lead to fewer breakdowns. In this model, the total robot downtime is minimized to get the optimum number of inspections.
The total robot downtime, TRDT, per unit time is defined as:
n is the number of inspections per robot facility per unit time.
Tdp is the downtime per inspection for a robot facility.
Tdb is the downtime per breakdown for a robot facility.
k is a constant for a specific robot facility
2.6.4 Reliablity Analysis of a Repairable/Non-Repairabe Robot System
This model represents a robot system composed of a robot and a safety unit. In the industrial sector, the inclusion of safety units or systems with robots is often practiced because of robot accidents involving humans. In this model, it is assumed that after the failure of the safety unit, the robot may fail safely or with an incident. The failed safety unit is repaired. The robot system state space diagram is shown in Fig 5. The numerals in boxes and circles denote system states
Fig 5. Robot system state space diagram
The following assumptions are associated with this model:
All failures are statistically independent.
All failure and repair rates are constant.
The robot system fails when the robot fails.
The repaired safety unit is as good as new.
The following symbols are associated with the diagram in Figure 5 and its associated equations:
Pi (t) is the probability that the robot system is in state i at time t; for i = 0 (robot and safety unit working normally), i = 1 (robot working normally, safety unit failed), i = 2 (robot failed with an incident), i = 3 (robot failed safely), i = 4 (robot failed, safety unit operating normally).
Î»rb is the robot failure rate.
Î»s is the safety unit failure rate.
Î»rbi is the rate of the robot failing with an incident.
Î»rbs is the rate of the robot failing safely.
Î¸s is the safety unit repair rate.
The robot system reliability (i. e., when both robot and safety unit work normally)with safety unit repair facility is given by :
Robot system mean time to failure is given as:
MTTFrbr is the robot system mean time to failure (i. e., when both robot and safety unit are working) with safety unit repair facility.
Robots are complex and sophisticated machines with good reliability. However, these machines do fail occasionally and thereby require maintenance similar to other engineering systems. Users of the robots, therefore, requires a sound maintenance program which would be beneficial in overcoming unscheduled down time and also increasing the robots efficiency. A poor maintenance program may also lead to safety-related problems.
Maintenance is required and helpful not only in the operational phase of robots but also during the design phase. Many decisions during the design phase are made regarding the maintenance of robots.
3.2 Robot Maintenance Requirements and Types :
The maintenance requirements of a robot are determined by the robot type and its application. Probably the most important part that affects the need for maintenance and the provision of maintenance is the robot power system .
Most robots used in the industrial sector can be grouped under two classiï¬cations:
(a) electrical and
(b) hydraulic with electrical controls .
Robots driven and controlled by electrically driven parts are referred to as electrical robots, whereas, robots whose working components are driven hydraulically and normally are referred to as hydraulic robots. Irrespective of robot type, the mechanical components of robots require careful attention.
Maintenance of robots used in the industrial sector can be divided into three basic categories as shown in Fig.6 These categories are preventive maintenance, corrective maintenance, and predictive maintenance.
Preventive maintenance is concerned with the periodic servicing of robot system components.
Fig 6 Basic types of maintenance for industrial robots.
Corrective maintenance is concerned with repairing the robot to an operational state after its breakdown.
Predictive maintenance is concerned with predicting failures that may occur and alerting the appropriate maintenance personnel. Many robots are equipped with sophisticated electronic components and sensors.
3.3 Robot Parts and Tools for Maintenance and Repair
A robot is composed of various parts, subsystems, and accessories. Maintenance personnel must be familiar with such robot elements in order to perform their tasks effectively which are as follows :
Hydraulic power supply
Cartesian coordinate system
Cathode ray tube (CRT)
Printed circuit board
Mass memory device
Strain gauge sensor
Alpha numeric keyboard
In robot maintenance various types of tools are used, ranging from wrenches to diagnostic codes displayed on the robot control panel. Although the maintenance tools required are peculiar to the specific robot system in question, some of the most commonly used tools are as follows :
Circuit card pullers
Accumulator charging adaptors
3.4 Robot Inspection
Usually, robots are inspected regularly by their users. Nonetheless, the inspections of industrial robots may be grouped into two broad categories.
Fig 7: Two broad categories of industrial robot inspections.
In category I some of the items checked prior to daily operations of the robots are:
The presence of abnormal noise
The presence of abnormality in the robot supply air pressure
The presence of abnormality in the robot supply oil pressure
The presence of abnormal vibrations
The presence of abnormality in the supply voltage
The proper working of interlocking between the contact prevention equipment and the robot
Damage to external electric wires and piping
The presence of abnormality in the robot operation
The proper working of interlocking the mechanism of associated items with the robot
The state of items used for the prevention of contact with the robot in operation
In category II some of the items checked at regular intervals are:
The looseness of bolts in major robot parts
Abnormal conditions in the electrical system
Abnormality in the servo-system
Abnormality in the power train
Abnormal conditions in stoppers
Abnormality in the operational troubleshooting function
Abnormality in the lubrication of movable parts
Abnormality in the air pressure system
Abnormality in the oil pressure system
3.5 Guidelines for Safeguarding Robot Maintenance personnel
During robot maintenance utmost care must be given to protect robot maintenance and repair personnel. Four useful guidelines for this purpose are to :
Ensure that all maintenance personnel are properly protected from unexpected robot motion
Ensure that all maintenance personnel have proper training in procedures appropriate to perform the required tasks safely
Ensure that when a lockout or tag out procedure is not used, equally effective alternative safeguarding methods are employed.
Ensure that the robot system is properly switched off during maintenance and repair activities as well as that the sources of power and the releasing of potentially dangerous stored energy are properly locked out or tagged
When it is not possible to turn off power during maintenance, some useful guidelines for protecting maintenance personnel are as follows :
Reduce the robot speed to a slow speed level.
Place the robot arm in a predetermined position so the required maintenance tasks can be performed without exposing humans to trapping points.
Make the emergency stop readily accessible and make restarting the robot impossible until the emergency stop device is reset through manual means.
Place the entire control of the robot in the hands of the maintenance person.
Place robot controls in the hands of a second person who is knowledgeable regarding potential robot-associated hazards and is capable of reacting fast to protect others in a moment of need.
Use devices such as pins and blocks during maintenance to prevent the robot system's potentially hazardous movements.
Failure Data Collection and Analysis
Failure data are the backbone of reliability studies because they provide invaluable information to concerned professionals such as reliability engineers, design engineers, and managers. In fact, the failure data of a product are the final proof of reliability related efforts expended during its design and manufacture. It would be impossible to have an effective reliability program without the collection, analysis, and use of information acquired through the testing and operation of products used in industrial, military, and consumer sectors.
The basic purpose of developing a formal system of collecting, analyzing, and retrieving information acquired from past experiences is to allow design and development of better and more reliable products without repeating the previous efforts concerning research, design, and testing to achieve the current product reliability. Thus, the fundamental goal of a failure data
collection and analysis system is to convert the relevant information accumulated in various sources into an effectively organized form so that it can efficiently be used by individuals with confidence in conducting their assigned reliability related tasks.
4.2 Uses of Failure Data with respect to Robots
Failure data plays an important role in various studies related to robots. Some of the areas in which failure data becomes useful are:
Estimating the robots or its components reliability, availability, and hazard rate or the failure rate.
Determining the maintenance requirements of a robot and its components
Conducting robot life-cycle cost studies
Gaining knowledge regarding robot design and manufacture deficiencies
Supporting robot reliability growth programs
Performing effective design reviews
Conducting cost-reliability trade off studies
Recommending design changes for improving the robot's or its part's reliability
4.3 Failure Reporting and Documentation System for Robots
In order to produce reliable and safe robots, establishment of reporting and documentation systems is necessary. Whenever such systems are being designed, attention must be paid to the following factors: 
Involvement of system users during design
Failure reporting form simplicity and clarity
Elimination of the need to memorize codes
Prerecording static information on the data form whenever feasible
Ease of using recorded data for analysis
4.4 Reliability Data Sources
During the product life cycle, there are several identifiable data sources, and some of these are tests (which can be environmental qualification, field installation, and field demonstration), warranty claims, previous experience with similar products, data systems operated by customers, repair records, inspection records, etc. As electrical, mechanical, and electronic components may be used in a robot, the failure data for such components could be obtained from the existing data banks in the various sector of the industry.
Field failure data collection
Major reliability data sources
Event data banks
Major reliability data sources are shown in the fig above:
Field data collection
Event data banks
Published information involves collecting generic failure data from open published literature e.g., conference proceedings, books, journals, industrial reports etc. Field data collection makes use of the information available in maintenance and operating records to develop an as-complete-as-possible historical records of events. Failure data is derived from this history. Laboratory tests involves grouping of similar items which are tested under laboratory conditions and time intervals are recorded with the failure events. Needed failure parameters are obtained from such information. In expert opinion, failure data are generated from the opinions of experts such as equipment operators and maintenance personnel which are then converted into quantified data using the paired comparisons, Delphi etc.
4.5 Hazard Plotting Method
This graphical method is used to determine a statistical distribution and its parameters from given failure data. The method was developed to handle multiple censored failure data composed of times to failure on failed items and running times on items which have not failed. The life data are called multiply censored when un failed items or units have distinct running times. On the other hand, data are called singly censored when all un failed items have the same censoring time and are larger than the failed unit's failure times. Situations such as removal of items from use before failure, collection of data while items are still functioning , and failure of items due to extraneous causes lead to multiply censored data. The failure of all units results in complete data.
Some of the benefits of this method are:
It fits the failure data to a straight line
It provides a complete and easily comprehensible picture of failure data under study
It can easily handle complete and incomplete failure data
It can easily be understood by persons with a less statistical background.
The hazard plot can be used to obtain information such as:
Estimates for distribution parameters
Estimates for percentiles of distribution
Individual failure modes life distributions
Percentage of items failing by a specific period
Resulting life distribution when specific failure modes are eliminated.
4.5.1 Steps involved in Hazard Plotting
Following steps are associated with hazard plotting :
Obtain failure/running time data
Identify running times with an asterisk (censoring times), when applicable
Order the failure and running times from smallest to largest
Associate with each failure time the reverse rank beginning with the size of data sample (number of data points)
For each failure compute hazard value
For each failure compute the cumulative hazard value
Select a statistical distribution and obtain its cumulative hazard function
Examine visually the plotted data points for straightness. If they reasonably follow a straight line, conclude that the assumed theoretical distribution adequately fits the data, and obtain values for the distribution parameters, If not , try another theoretical distribution.
This report emphasizes that reliability, maintenance and safety of robots are inter related issues that requires substantial input and commitment from all the parties involved ranging from manufacturer to the end user .
Manufacturer has to play an important role by going through all the stages including the robot performance testing, robot reliability testing and providing the proper specification and information on robots and proper design solutions thereby ensuring the reliability and safety of robots. End users, the industry, on the other hand, can influence the effectiveness of robot systems following preventive, corrective and predictive maintenance procedures and also making sure that the guidelines for safeguarding robot maintenance personnel are followed properly thereby increasing the productivity and the effectiveness of robots.
Failure data analysis also plays an important role by estimating the robots reliability, availability, and hazard rate, determining the maintenance requirements and manufacture deficiencies, and recommending design changes in order to improve the reliability.
Robot has become a necessity these days ranging from a house hold to complex military purposes and space expeditions to healthcare industry thereby emphasizing the importance of robot reliability, safety and quality. The subject of robot reliability is very complex and there are numerous interlocking variables in evaluating and accomplishing various reliability levels. A successful robot installation has to be safe and reliable. A robot with poor reliability leads to many problems such as high maintenance cost, unsafe conditions and inconvenience.
Robot reliability depends on four factors( design, manufacturing, environment and user), hence, in order to make a robot reliable we should keep these four factors in our mind which are directly linked to the reliability of the robot.