Table of Contents:
Robots are increasingly being used for performing various types of tasks including arc welding, spot welding, routing, and materials handling. A robot may simply be described as a mechanism guided by automatic controls and the word “robot” is derived from the Czechoslovakian language, in which it means “worker” .
The first commercial robot was manufactured by the Planet Corporation in 1959 . Nowadays, millions of industrial robots are being used around the globe . As robots use electrical, mechanical, hydraulic, pneumatic, and electronic components, their reliability-associated problems are highly challenging because of many different sources of failures. Although there is no clear-cut definitive point in the beginning of robot reliability field, a publication by JF Engelberger, in 1974, could be regarded as its starting point . A comprehensive list of publications on robot reliability is available in Refs. [5,6].
This chapter presents various important aspects of robot reliability.
Robot Failure Classifications, Causes, and Corrective Measures
Robot failures can be categorized under the four classifications as shown in Figure 5.1 [7-9].
Classification I (i.e., human errors) failures occur due to the personnel who design, manufacture, test, operate, and maintain robots. Some of the causes for the occurrence of human errors are as follows: 
FIGURE 5.1 Robot failure classifications.
Thus, human errors may be divided into categories such as inspection errors, design errors, installation errors, operating errors, assembly errors, and maintenance errors. Some of the methods that can be used to reduce the occurrence of human errors are as follows:
The first method (i.e., fault tree analysis) is described in Chapter 4 and the remaining three methods are described in Dhillon .
Classification II (i.e., software failures/errors) failures are associated with robot software and in robots, software faults/failures/errors can happen in the embedded software or the controlling software and application software. Redundancy, even though it is expensive, is probably the best solution to protect against the occurrence of software failures/errors. Also, the application of approaches such as fault tree analysis, failure modes and effect analysis, and testing can be quite useful for reducing the occurrence of software failures/errors. Furthermore, there are many software reliability models that can also be used for evaluating reliability when the software in question is put into operational use [11-13].
Classification III (i.e., systematic hardware faults) failures are those failures that can take place due to unrevealed mechanisms present in the robot system design. Some of the reasons for the occurrence of such faults are unusual joint-to-straight-line mode transition, failure to make the appropriate environment-associated provisions in the initial design, and peculiar wrist orientations.
Some of the methods that can be employed for reducing the occurrence of robot- related systematic hardware failures are the use of sensors for detecting the loss of pneumatic pressure, line voltage, or hydraulic pressure; and the employment of sensors for detecting excessiveness of temperature, acceleration, speed, force, and servo errors. Several methods considered quite useful for reducing systematic hardware failures are described in Dhillon .
Finally, Classification IV (i.e., random component failures) failures are those failures that occur unpredictably during the useful life of components. Some of the reasons for the occurrence of such failures are undetectable defects, unavoidable failures, low safety factors, and unexplainable causes. The methods presented in Chapter 4 can be used for reducing the occurrence of such failures.
Robot Reliability–Related Survey Results and Robot Effectiveness Dictating Factors
Jones and Daw'son  reported the results of a robot reliability study that was based on surveys of 37 robots of 4 different design used in 3 different companies X, Y, and Z; covering 21,932 robot production hours. These three companies (i.e., X, Y, and Z) reported 47, 306, and 155 cases of robot reliability-related problems, respectively, of which the corresponding 27, 35, and 1 cases did not contribute to any downtime. More specifically, robot downtime as a proportion of production time for these three companies (i.e., X, Y, and Z) was 1.8%, 13.6%, and 5.1, respectively.
Approximate mean time to robot failure (MTTRF) and mean time to robot-related problems (MTTRP) in hours for companies X, Y, and Z are shown in Figure 5.2.
It is to be noted that as shown in Figure 5.2, among these three companies, there is a w'ide variation of MTTRF and MTTRR More specifically, the highest to lowest MTTRF and MTTRP levels are 2596.40 and 221.15 hours, respectively.
FIGURE 5.2 Approximate mean time to robot failure (MTTRF) and mean time to robot- related problems (MTTRP) in hours (h) for companies X, Y. and Z
There are many factors that dictate the effectiveness of robots. Some of these factors are as follows [9,15]: