Jump to ContentJump to Main Navigation
Show Summary Details
More options …

Journal of Intelligent Systems

Editor-in-Chief: Fleyeh, Hasan


CiteScore 2017: 0.96

SCImago Journal Rank (SJR) 2017: 0.193
Source Normalized Impact per Paper (SNIP) 2017: 0.481

Online
ISSN
2191-026X
See all formats and pricing
More options …
Volume 24, Issue 1

Issues

Empirically Investigating a Hybrid Lean-Agile Design Paradigm for Mobile Robots

Salah A.M. Elmoselhy
Published Online: 2014-08-20 | DOI: https://doi.org/10.1515/jisys-2014-0024

Abstract

Lean design and agile design paradigms have been proposed for designing robots; yet, none of them could strike a balance between cost-effectiveness and short duration of the design process without compromising the quality of performance. The present article identifies the key determinants of the mobile robots development process. It also identifies empirically the mobile robot design activities and strategies with the most influence on mobile robot performance. The study identified statistically the mobile robot design activities and strategies most positively correlated with mobile robot performance. The results showed that 65% of typical mobile robot design activities and strategies are affiliated with the lean design paradigm, while the remaining 35% are affiliated with the agile design paradigm. In addition, it was found that 22% of the lean mobile robot design activities and strategies and 25% of the agile mobile robot design activities and strategies, significantly with 99% confidence, are among the design activities and strategies most positively correlated with improving mobile robot performance. A hybrid lean-agile design paradigm is thus proposed.

Keywords: Robotics; mechanisms; robots; robot dynamics and control; applications of design theory; design techniques; operations research and management science; MSC 2010 68T40; 70B15; 70E60; 94C30; 93B51; 90Bxx

1 Introduction

Lean and agile approaches have been adopted by designers of robots for years. Lean design involves value optimization through minimizing waste in the design process [7]. Lean design thus usually leads to cost reduction. According to Womack and colleagues [40, 41], significant interest has been shown in recent years in the idea of lean operations. Efficient resources allocation is a key aspect for minimizing waste [38]. More recently, a growing awareness has been established that lean principles can be readily transferred to the design sector [7]. The lean design process would only be successful if the success criterion was only cost. Agile design is a design system with flexible technology, qualified and trained human resources, and shared information that responds quickly to continuous and unpredicted changes in customers’ needs and desires and in market demand [43]. Having this ability can make the mobile robot design process successful if the success criterion is short lead time [19, 31].

The current challenge in the design process of robots is to improve the value added to customer while shortening the robot design process duration [5, 10]. Therefore, robot designers face a dilemma that they need to strike a balance between the robot design duration and the robot performance in the most cost-effective way. Recently, Chalupnik et al. [15, 16] extensively investigated minimizing variations in performance caused by variations in uncontrollable external noise parameters or by variations in design parameters. More recently, Aravinth et al. [6] proposed minimizing variations in performance caused by variations in internal factors in the design process of complex products, such as failure modes. Bao et al. [9] also investigated more recently the optimization of task allocation in collaborative customized product development. Yet, internal factors in the robot design process, such as design activities and strategies, and their relation to robot performance have not yet been investigated empirically. Browning et al. [12] reported that determining how and when value is added in the design process is problematic. Thus, the present research aims to help novel designers of mobile robots in resolving this dilemma by identifying empirically the mobile robot design activities and strategies with the most influence on robot performance and by identifying a potentially efficient mobile robot design paradigm.

This research investigates how and when value is added in the design process of mobile robots. The article starts with identifying the key determinants of the mobile robots development process. The technical attributes and design specifications of mobile robots are then presented. A quasi-experiment on the design of mobile robots is presented after that as a case study. Following from this experiment, the design activities and strategies typically implemented in the mobile robot design process and their lean-agile classification are investigated next. The study then identifies the lean mobile robot design activities and strategies and agile mobile robot design activities and strategies with the most influence on mobile robot performance. Finally, the article investigates a potentially efficient design paradigm in the design process of mobile robots.

2 Key Determinants of the Mobile Robots Development Process

Design managers in industry face a dilemma that they need to strike a balance between the product development time and the product performance attributes cost-effectively. The available literature on the mobile robots development process has identified this dilemma but has not as yet resolved it. This research helps in resolving this dilemma by identifying the key determinants of the mobile robots development process that primarily drive and shape the design process and by which the design process outcome is mainly determined. It was found in this study that there are primarily five determinants of the mobile robots development process, which are shown in Figure 1.

Key Determinants of the Mobile Robots Development Process.
Figure 1

Key Determinants of the Mobile Robots Development Process.

As Figure 1 depicts, the key determinants of the mobile robots development process are the customers’ perceived values and the strategic business goals of the mobile robot development organization, which collectively through the mobile robot development time duration and mobile robot development cost shape the mobile robot attributes. These five determinants are interconnected through the mobile robot development activities and strategies. The research findings indicated in Figure 1 are backed by the research findings of Cohen et al. [21], in which they indicated that product development time, cost, and attributes are among the key product development metrics. The other two identified key determinants of the mobile robots development process, i.e., customers’ perceived values and strategic business goals of the mobile robot development organization, are also backed by the research findings of Eschenbaecher and Graser [24]. These key determinants have to be harmoniously integrated if we are to make the mobile robots development process a success business-wise. The mobile robot attributes are set industrially on the basis of customers’ perceived values and on SWOT (i.e., strengths, weaknesses, opportunities, and threats) analysis. Strategic business goals are based on SWOT analysis. To optimize mobile robot development cost and development time duration, an FMEA (i.e., failure modes and effects analysis) is conducted in the mobile robots development process. This being a case study, a mobile robot design quasi-experiment was conducted.

3 Mobile Robots Performance and Technical Attributes

Mobile robots are industrially sought for their advantages that range from reducing operating costs, improving product quality and consistency, as well as the quality of work for employees, to increasing production output rates, increasing product manufacturing flexibility, reducing raw materials waste, and increasing yield [1]. A recent trend in robots design is mobile robots [17]; thus, the design of mobile robots was chosen to be the basis of the design experiment in the present research. The quality of mobile robots is measured against the following mobile robot performance attributes that were used as the rubric of evaluation in this design experiment and that were extracted from the industrially adopted set of technical attributes of a quality mobile robot [1, 25]: (i) minimum floor space requirements for agile motion; (ii) adaptable to the surroundings and capable of making decisions accordingly, such as in case of path irregularity, missing a junction, and encountering obstacles [3]; (iii) capable of recognizing the position of an object; (iv) capable of controlling the force used to grip an object; (v) provides flexibility for picking and depositing loads to a variety of station types and elevations; (vi) capable of following a non-straightforward path; (vii) fast response [8]; (viii) stability [8, 32]; (ix) accuracy [8]; (x) payload capacity; (xi) reliability; (xii) maintainability; and (xiii) safety. The degree of striking a balance between these competing technical attributes shapes the mobile robot performance, which in turn should be realized cost-effectively within the shortest duration of robot design process possible [30, 33]. These attributes have been observed in the mobile robot design quasi-experiment.

4 Mobile Robots Design Quasi-experiment

A quasi-experiment on mobile robot design was conducted on the basis of a contest among groups of novel designers who were undergraduates at the Engineering Department of Cambridge University to design, build, and test a mobile robot such that the robot carries out a set of tasks successfully within a certain time frame. In this experiment, the novel designers were observed while designing, building, and testing their mobile robots, and were asked to respond to a questionnaire on the design activities and strategies they adopted. The experiment took a month to complete. The designers were free to choose their design approach. It was not a prerequisite for the designers to attend a course on agile or lean design. The design-independent variables in this experiment were the design activities and strategies, and the design-dependent variable was the robot performance. The data collected for analysis were from observations and from the responses to the questionnaire. The following subsections elaborate on the mobile robot design experimental setup and specification, mobile robot performance evaluation criteria, method of analysis of the experimental results, and experimental observations and responses to the questionnaire that all have been adopted in this mobile robot design experiment.

4.1 Mobile Robots Design Experimental Setup and Specification

In the experimental setup of this quasi-experiment, each design team was divided into three subteams: one for the mechanical aspects, one for the electronics aspects, and one for the software aspects. The aim of the design experiment was to design and build a mobile robot that is able to collect six pallets from a conveyor belt, “B,” and to transport them to one of two delivery points, “D1” or “D2,” depending on the type of pallet within 5 min, as illustrated in Figure 2 [14]. The task would continue until half-a-dozen pallets were transferred or the time limit was reached. In this experiment, the following conditions were applied: (i) the conveyer that is indicated in light green in Figure 2 could be started/stopped and reversed using light-dependent resistor optical switches; (ii) an adjustable light-emitting diode-based beam suitable for driving an optical sensor, mounted just above the conveyer belt, was available [14].

Mobile Robot Design Experiment Contest Area Topography [14].
Figure 2

Mobile Robot Design Experiment Contest Area Topography [14].

Concerning the mechanical subsystem in this experiment, there was a set of resources available for the competing design groups. These resources available for the mechanical subsystem included (i) transmission components such as wheels, (ii) castors, (iii) D.C. motors, (iv) gearboxes, (v) pneumatic actuators, (vi) swivel connector, (vii) pneumatic valve assembly, (viii) pneumatic hoses and connectors, (ix) fasteners, (x) springs, (xi) spur and bevel gears, (xii) gear racks, (xiii) bearings, (xiv) adhesives, (xv) lubricants, and (xvi) structural materials with the availability of a workshop for processing these structural materials [14].

The competing design groups in the domain of electronics subsystem were also provided with a set of resources. These resources for the electronics subsystem included (i) an I2C bus, (ii) transducers, (iii) LEDs, (iv) ICs, (v) diodes, (vi) MOSFETs, (vii) capacitors, (viii) resistors, (ix) infrared emitter/detector assemblies, (x) infrared detector amplifier assembly, (xi) potentiometers, (xii) D plugs, (xiii) a data sheet providing data on I/O ports of the D.C. motors, (xiv) PCBs, (xv) soldering and circuit construction equipment, (xvi) 5 V power supply lead for prototyping, (xvii) motor/gearbox to PCB header lead, (xviii) microcontroller to PCB 12 V lead, and (xix) thermistor assembly and thermocouple materials [14].

Concerning the software subsystem, there was a set of resources available for the competing design groups. These resources for the software subsystem included a (i) C++ compiler, (ii) 32-bit microcontroller, (iii) software library, and (iv) power supply unit plus output lead [14].

In the simulation setup, the competing design groups in the domain of mechanical subsystem were provided with computers, a CAD system, and a CAM system. The competing design groups in the domain of software subsystem were provided with computers, sensor simulation PCB, and I2C cable [14]. Having seen this, let us now investigate the performance evaluation criteria adopted in this design experiment.

4.2 Evaluation Criteria for the Mobile Robot Design Experimental Performance

The robot performance evaluation criteria are collectively the mobile robot performance score based on conducting a set of tasks, accomplishing which needs meeting the mobile robot performance attributes mentioned in Section 2, within a specific time frame. The task to be performed by the robot was to collect six pallets from a conveyor belt, “B,” and to transport them to one of two delivery points, “D1” or “D2,” as illustrated in Figure 2, depending on the type of pallet, within 5 min. The task would continue until half-a-dozen pallets were transferred or the time limit was reached. Now, let us have a look at the method of analysis of the experimental results.

4.3 Method of Analysis of the Mobile Robot Design Experimental Results

The analysis of the mobile robot design experimental results in this research uses both descriptive statistics and inferential statistics. In the inferential statistics part, the non-parametric-statistic tool in the Statistical Package for Social Sciences was used to avoid making assumptions about the population’s parameters and consequently to improve the validity of the statistical analysis results. To identify which design team implemented which design activity and strategy to what extent in the experiment, a questionnaire listing the design activities and strategies, and the degree of their implementation, was constructed. To investigate the possible binding between some design activities and strategies, a dependency analysis was conducted. The assessment of how effective was the implementation of the design activities and strategies is implied in the correlation between the extent of their implementation and the performance score given by the design team. The adopted statistical approach has the following five attributes: (i) the population consists of novel designers of mobile robots; (ii) the sampling frame is based on a bounded and unbiased collection of designers in which every single individual is identified and can be selected; (iii) the data type is a categorical random variable; (iv) the sampling design is based on the probability simple random sampling because of its cost-effectiveness and reasonable accuracy; and (v) the target sample size is not less than 30 novel designers, which is the minimum statistically representative sample size [2, 39]. This method of assessment has been applied to the experimental observations and responses to the questionnaire.

4.4 Experimental Observations and Responses to the Questionnaire

The research observations obtained from observing the mobile robot design activities in the experiment were further verified by including them in the questionnaire handed to the novel designers. The questionnaire was administered to the respondents in 2008. The responses to the questionnaire were categorized as follows: “strongly disagree” is ranked 1, “disagree” is ranked 2, “agree” is ranked 3, and “strongly agree” is ranked 4. An average value in each column is used to fill in the gap of empty responses as a way of manipulating responses to questionnaires [37].

There were 29 design teams participating in this design experiment and a sample size of 174 could be realized, which exceeds the requirement of the minimum representative sample size of 30 novel designers. This satisfies the first of the two criteria for the sample attribute of being representative, which are sample size and sampling design. The second criterion for the sample attribute of being representative, i.e., sampling design, was also satisfied as the sampling design in this research was based on probability simple random sampling, which is suitable for limited generalization purposes and cost-effectively leads to reasonably fair results. The pragmatic reader might now well ask, “How have the results been analyzed?” The next section will answer this question.

5 Statistical Analysis of the Results

The observations in this mobile robot design experiment showed that there were design strategies and activities that were commonly adopted by the entire mobile robots novel designers involved in the experiment, and there were some mobile robot design strategies and activities that were adopted only by some of them. The scores achieved by the design teams showed that for six design teams, the performance of their robots was superior. A statistical analysis was conducted for each design team to examine whether there is a correlation between their mobile robot performance and the design activities and strategies they adopted. In the descriptive statistics, a frequency analysis of the data was conducted, including the mean and standard deviation. In the inferential analysis, a non-parametric statistical analysis was conducted using Spearman correlation coefficient, which provides more rigorous results than the parametric statistical analysis [39]. This section will present, first, the results of analysis of bivariate correlation with robot performance. Second, frequency analysis of the data is presented. Third, the results of the non-parametric statistical analysis and the results of the dependency analysis are demonstrated. Thereafter, the reliability analysis results are presented. Finally, the implication of the percentage of variation in robot performance due to a design variable (r2) is demonstrated.

5.1 Bivariate Correlation between Lean and Agile Design Activities and Strategies and Mobile Robot Performance

The ranges of statistical correlation adopted in the present research are as follows: (i) no correlation when the correlation coefficient ranges from 0 to <0.1; (ii) low correlation when the correlation coefficient ranges from 0.1 to <0.3; (iii) moderate correlation when the correlation coefficient ranges from 0.3 to <0.6; and (iv) high correlation when the correlation coefficient ranges from 0.6 to 1 [20]. This section presents the results related to the lean mobile robot design activities and strategies, and the agile mobile robot design activities and strategies. It was found that 65% of the total mobile robot design activities and strategies are affiliated with the lean design paradigm. It was also found with 99% confidence that 22% of these lean robot design activities and strategies significantly are among the most positively correlated design activities and strategies with improving mobile robot performance; these particular mobile robot design activities and strategies have been proved to significantly improve mobile robot performance by >10%. These lean design activities and strategies positively correlated with mobile robot performance are presented in Appendix A.

It was also found that 35% of the total mobile robot design activities and strategies are affiliated with the agile design paradigm. Moreover, it was found with 99% confidence that 25% of these agile robot design activities and strategies significantly are among the most positively correlated design activities and strategies with improving mobile robot performance; these particular mobile robot design activities and strategies have been proved to significantly improve mobile robot performance by >10%. These agile design activities and strategies positively correlated with mobile robot performance are presented in Appendix B. Examples of the questions used in the questionnaire are provided in Appendix C. A key reason for having more lean than agile design strategies/activities is that the lean design paradigm is more basically related to the key determinants of the mobile robots development process indicated in Figure 1 than the agile design paradigm. In addition, lean design strategies are basically more intuitive to designers than agile design strategies.

The appendices show the lean-agile design activities and strategies that are most influential on mobile robot performance, sorted in descending order, respectively. In the appendices, the first column entitled “Observation/Hypothesis” presents a brief description of the design activity/strategy. The next column entitled “Design Phase” implies to which design phase, i.e., scope-based, conceptual, preliminary, and detailed design phases, the investigated design activity/strategy is related. The third column entitled “Design Strategy/Activity” shows whether the design variable under investigation is a design strategy or design activity. The fourth column entitled “Observation/Hypothesis” elucidates whether the design variable under investigation was identified due to experimental observation or by a hypothesis deduced from literature review. These observations were made a priori before the questionnaire was handed to the designers. The term “observation” hence implies the description of what the designers have already done in this design experiment. The fifth column in the appendices entitled “Reference in Literature to Hypothesis” refers to the relevant references in the literature for those design variables that have been identified through a literature review. The last column entitled “Percentage of Variation in Mobile Robot Performance (r2)” depicts the corresponding value of r2 for each design variable under investigation. Having seen this, let us now investigate the frequency analysis of the data.

5.2 Frequency Analysis of Data

Frequency analysis provides an insight into the descriptive statistics of the collected data and of the categories of the collected data. This section presents the frequency analysis of the design activity most positively correlated with mobile robot performance. The design activity most positively correlated with mobile robot performance was to have the largest number of design iterations, if any, to occur within the software subsystem, i.e., agile design activity #1 in Appendix B. This section shows the frequencies and descriptive statistics of this design activity with mobile robot performance. Table 1 shows that the total valid percentage of data was 99.4%, which is proof of valid results.

Table 1

Frequencies and Descriptive Statistics of the Design Activity of Having the Largest Number of Iterations in the Software Subsystem.

Table 1 also shows that the largest percentage of responses as to this design activity, i.e., 41.4%, was of the “agree” category. This section paves the way to investigate whether there was an interdependency among the data.

5.3 Non-parametric Analysis and Dependency Analysis Results Using Bivariate Correlation and Partial Correlation Analyses

To investigate how rigorous the bivariate correlation analysis results are, a dependency analysis was conducted. A dependency analysis explores whether an independent variable, which was proven to be correlated with a dependent variable, is in turn a dependent variable on other variables. The dependency analysis in the present study was two-fold. First, a mutual dependency analysis based on a bivariate correlation coefficient was conducted. Second, a partial correlation analysis was conducted consequently to control for the effect of each of the two mutually dependent variables on each other in relation to other variables.

It was found that there was only a pair of mobile robot design activities and strategies that has mutual dependency. This pair consists of the lean mobile robot design strategy, which is to adopt a modular design, and the lean mobile robot design strategy, which is to strike a balance between functionality and design iterations. The bivariate correlation coefficient between these two design strategies was 0.711, as shown in Table 2, which plainly implies a strong potential for mutual dependency, as it was >0.6 [20]. Hence, a partial correlation analysis between the lean mobile robot design strategy of adopting modular design and mobile robot performance, controlling for the lean mobile robot design strategy of striking a balance between functionality and design iterations, was conducted, as shown in Table 3.

Table 2

Result of the Dependency Analysis – Bivariate Correlation.

Table 3

Partial Correlation between the Lean Mobile Robot Design Strategy of Adopting Modular Design and Normalized Performance Score.

The results of the partial correlation analysis showed that the effect of adopting the design strategy of striking a balance between functionality and design iterations on the relation between the design strategy of modular design and mobile robot performance was 0.049, as indicated in Table 3. Plainly, this value of partial correlation is negligible as the amount of influence was <5% [20]. Hence, the correlation coefficient between the design strategy of modular design and robot performance remained unchanged in the low correlation category. Also, a partial correlation analysis between the design strategy of striking a balance between functionality and design iterations and mobile robot performance, controlling for the design strategy of modular design, was conducted, as shown in Table 4. In Tables 3 and 4, the degree of freedom (df) indicates the valid number of participating designers.

Table 4

Partial Correlation between the Lean Mobile Robot Design Strategy of Striking a Balance between Functionality and Design Iterations and Normalized Performance Score.

The results of the partial correlation analysis showed that the effect of the design strategy of modular design on the relation between the design strategy of striking a balance between functionality and design iterations and mobile robot performance was 0.009, as indicated in Table 4. This value of partial correlation is also plainly negligible as the amount of influence was <5% [20]. Thus, the correlation coefficient between the design strategy of striking a balance between functionality and design iterations and mobile robot performance remained unchanged in the no correlation category. This might raise the following question: “How reliable are these results?” This question will be addressed in the following section.

5.4 Reliability Analysis of Results

In reliability statistics, if the data collected reached the level of ≥0.7 on the Cronbach’s α scale, then the collected data have good internal consistency [37]. As the Cronbach’s α internal reliability factor of the collected data was 0.705, as shown in Table 5, the data have good internal consistency, based on the average inter-item correlation.

Table 5

Reliability Statistics.

There are a couple of assumptions based on which the reliability analysis was conducted: (i) observations are independent; (ii) errors are uncorrelated between the mobile robot design activities and strategies. The following section explores the implication of these results.

5.5 Implication of Percentage of Variation in Mobile Robot Performance due to a Design Variable (r2)

In a research project that includes several variables, it is often sought to know how one variable is related to another. Correlational research attempts to determine whether, and to what degree, a relation exists between two or more quantifiable variables, such as two design activities. Correlation implies prediction of the value of one variable if we know the value of the other correlated variable, but does not necessarily imply full causation. The reason why correlation does not necessarily imply full causality is that a third variable may be involved of which we are not aware. However, correlational research can imply partial correlation in terms of prediction of percentage of variation in, for instance, variable “B” due to variable “A” [20].

The correlation coefficient (r) value ranges from –1 to 1. Having correlation coefficient of a value “–1” indicates a perfect negative relation between the variables under examination. If the correlation coefficient has a value of “0,” there is no relation between the variables under examination. A correlation coefficient of a value “1” is interpreted as a perfect positive relation between the variables under examination. The square of the correlation coefficient (r2) represents the percentage of variation in one of the two variables under investigation due to the other correlated variable, which implies a causal link between these two variables.

Causality in this research was determined according to the percentage of variation in technical performance due to the variable (r2) that is based on the correlation coefficient (r). The design activities and strategies most correlated to mobile robot performance were determined according to their percentage of variation in technical performance due to the variable correlation coefficients with mobile robot performance, and according to their resulting p-value.

The aggregation of the percentage of variation in mobile robot performance due to the design variables reached collectively >100%, as there was an overlap in the affected areas in mobile robot performance by the design variables. This research shows that the design activities and strategies indicated in Appendices A and B, i.e., design variables, are independent except for 3% of them where a mutual dependency was identified as proved in the dependency analysis results indicated in Section 4.3. This helps now in deducing a potentially efficient mobile robot design paradigm.

6 Hybrid Lean-Agile Mobile Robot Design Process

On the basis of the presented results, adopting both lean robot design activities and strategies and agile robot design activities and strategies together in the mobile robot design process was proved to be practically valid. In addition, the design experiment proved that both lean and agile mobile robot design activities and strategies are correlated with and have significant influence on improving mobile robot performance. Besides, it was found that there are mobile robot design activities that have attributes of both the lean and agile design paradigms. For instance, the design activity of evaluating design concepts exhibits attributes of both lean and agile design paradigms. This further supports the practical validity of adopting both lean and agile mobile robot design activities and strategies in the mobile robot design process. Therefore, the present research proposes a hybrid lean-agile mobile robot design paradigm in which both lean mobile robot design and agile mobile robot design activities and strategies are adopted in the mobile robot design process benefiting from the attributes of both the lean design and agile design paradigms.

The proposed mobile robots hybrid lean-agile design pillars include (i) adopting the most effective lean design strategies, such as considering the reliability of the mobile robot in the design process in terms of the ability of the mobile robot to perform its required functions under stated conditions for a specified robot service time; (ii) adopting the most effective agile design strategies, such as having designs that are less vulnerable to failure modes and are less exposed to and less sensitive to the uncontrollable external factors by shifting complexity to the software subsystem rather than to the mechanical subsystem; (iii) adopting the most effective lean design activities, such as adopting testable design interdeliverables within and among system modules based on project milestones to detect mistakes as early as possible and to minimize the impact of mistakes on the successful completion of the design project; (iv) adopting the most effective agile design activities, such as having iterations in the software subsystem rather than in the mechanical subsystem to end up with a shorter development time; (v) adopting a three-phase hybrid lean-agile risk management action plan that helps in integrating mobile robot design activities and strategies to minimize risk in the mobile robot design process; and (vi) adopting a mobile robot design functional strategy in terms of the following items: standard components, modular design, communized architecture of mobile robots chassis and frame parts, and concurrent engineering in the design process.

The phases of the proposed three-phase hybrid lean-agile risk management action plan are as follows: (i) before the beginning of the mobile robot design process in which SWOT analysis is conducted; (ii) during the mobile robot design process in which the design team proves the value of the design concept to stakeholders at the end of each design phase, ensuring that the mobile robot satisfies stakeholders, fits its intended purpose, is of a quality to last its design lifetime, and can be made at an acceptable cost; (iii) after the end of the mobile robot design process in which an FMEA is conducted, and ultimately the models of mobile robots that fall short of the set target are killed off as soon as this appears. This approach to managing risk in the product design process is expected to help in realizing the sought harmonious integration between the product development activities and strategies.

Key challenges on the hybrid lean-agile mobile robot design paradigm include striking cost-effectively a balance between robot quality and short duration of the design process. In addition, misreporting cost savings can hurt the credibility of the hybrid lean-agile mobile robot design paradigm practitioners. Moreover, lean savings are indeed a long-term proposition. There is expectedly much pressure imposed on the implementers of the hybrid lean-agile mobile robot design paradigm to show immediate savings in terms of financial indicators such as the return on investment (ROI) to top management. Yet, reduction in defects and reduced cycle times are all areas that will continue to produce savings long after the term of ROI has run out. The implementation of the hybrid lean-agile mobile robot design paradigm thus needs a champion to lead cost-effectively the change and leverage it. The pragmatic reader is now invited to explore how valid this research is.

7 Results and Discussion

The questionnaire for collecting data for the present study was designed with emphasis placed on maximizing the clarity of the wording of the questionnaire and minimizing the influence of the questionnaire’s problems such as bias. To maximize the clarity of questions, clarification footnotes were used. In addition and in order to minimize bias, a two-fold strategy was adopted; first, in order to avoid researcher’s bias, closed-ended questions were used; second, in order to spot respondent’s bias, repeatedly inverted questions were used. The statistical sampling in this research is representative and the experimental results were statistically significant with 99% confidence, without making any assumptions about the population of novel designers of mobile robots.

In this study, the reliability statistics test results, based on Spearman correlation coefficient and non-parametric statistical analysis, have proved the reliability of the data used in this research and thus have verified our results. To investigate the validity of the results of this study, an inferential statistical analysis was conducted on the resulting correlation coefficients. The p-value was adopted as a measure that a result is true to the population. A cutoff p-value of 0.1 was adopted in this research. In addition, the statistical sampling in this research is representative in terms of sampling design that is suitable for limited generalization with cost-effectively fair statistical results and sample size that satisfies the minimum statistically representative sample size. Causality in this study was determined according to the percentage of variation in robot performance due to a variable (r2). The research results are also valid in terms of the four validity types: first, in terms of statistical conclusion validity, as the resulted relations are meaningful and reasonable; second, in terms of internal validity, as the results are causal rather than being just descriptive; third, in terms of construct validity, as the results represent what is theoretically intended; fourth, in terms of external validity, as the results can be limitedly generalized to the population of novel designers of mobile robots. Hence, this research has proved the validity of the correlation between lean and agile mobile robot design activities and strategies and mobile robot performance. In addition, it has validated the practicality of having both lean mobile robot design activities and strategies and agile mobile robot design activities and strategies implemented in the mobile robot design process.

The study showed that 65% of typical mobile robot design activities and strategies are affiliated with the lean design paradigm, while the remaining 35% are affiliated with the agile design paradigm. In addition, it was found with 99% confidence that 22% of the lean mobile robot design activities and strategies and 25% of the agile mobile robot design activities and strategies significantly are among the design activities and strategies most positively correlated with improving mobile robot performance. These particular mobile robot design activities and strategies proved to significantly improve mobile robot performance by >10% and thus should receive the highest priority of being assigned design process resources, such as those indicated in Section 4.1. Owing to the usually limited resources available for a design project to be done, there is a sort of trade-off and priority should be given to the design strategies/activities of higher correlation. The priority on resources allocation should be given to the design strategies/activities that have a correlation r2 value >0.1. A key reason why most of the design strategies/activities are weakly correlated with mobile robot performance is those particular strategies/activities are necessarily basic strategies/activities that are commonly shared among designers. The design strategies/activities that have a higher correlation are among the differentiators between the highly successful mobile robots and the rest of the mobile robots. As can be gathered from Appendices A and B, the higher correlation was for the conceptual, preliminary, and detailed design phases rather than the phase of design scope substantially because the conceptual, preliminary, and detailed design phases cover most of the time span of the design project and cover a larger number of strategies so that they have a higher probability to be more influential. A few design strategies/activities were expected to have higher correlation, such as the lean design strategy #6 in Appendix A. Thus, the study showed that with 99% confidence, >10% of the variation in mobile robot performance can be explained by adopting a hybrid lean-agile mobile robot design paradigm that adopts both lean and agile mobile robot design activities and strategies together in the mobile robot design process. Hence, adopting a hybrid lean-agile mobile robot design paradigm is technically valid.

8 Conclusion

This study determined how and when value is added in the mobile robot design process by identifying empirically the mobile robot design activities and strategies with the most influence on mobile robot performance. The article identified, first, the key determinants of the mobile robots development process and the key technical attributes of mobile robots. Second, the research method and statistical analysis helped in identifying the causal relations between the lean and agile mobile robot design activities and strategies and mobile robot performance, as presented in Appendices A and B. Finally, the research method and statistical analysis also helped in proving that adopting both lean and agile mobile robot design activities and strategies and strategies in the mobile robot design process is practically valid.

The study proposes the pillars of the mobile robots hybrid lean-agile design paradigm. It shows that with 99% confidence, >10% of the variation in mobile robot performance can be explained by adopting a hybrid lean-agile mobile robot design paradigm that adopts both lean mobile robot design activities and strategies and agile mobile robot design activities and strategies together in the mobile robot design process. Hence, adopting a hybrid lean-agile mobile robot design paradigm is technically valid. To further investigate the validity of these results, the study proposes the use of a larger sample size that is large enough to represent the whole population of novel designers of mobile robots. In addition, the study proposes the use of an industrial setting to further investigate the validity of these results.

Acknowledgments

The staff of the Cambridge Engineering Design Centre, Cambridge University, are thanked for their help in accomplishing this work. The financial support provided for this research by the EPSRC under IMRC grant number EP/E001777/1 and by Cambridge Overseas Trust is acknowledged.

Appendix A

Appendix A

Lean Activities and Strategies Positively Correlated with Mobile Robot Performance

Appendix B

Appendix B

Agile Activities and Strategies Positively Correlated with Mobile Robot Performance.

Appendix C

Mobile Robots Design Questionnaire

The aim of this questionnaire is to explore, for research purposes, the hypothesized correlation and causality between design process activities and strategies, on the one hand, and the quality of the final product performance. In order to explore this hypothesized correlation, we would like to collect some data from designers about their real design process activities and strategies. Consequently, we would like you to kindly answer this questionnaire. This questionnaire should be returned to Salah Elmoselhy (sae33@cam.ac.uk). Thank you in advance for your cooperation.

Please, tick (√) the most appropriate evaluation category to what your design team actually adopted and conducted in the design process.

Design Team Number:

Bibliography

  • [1]

    ABB Robotics Product Guide, Available at: http://www.abb.com/product/ap/seitp327/cc4949febe7dcfe9c12573fa0057007a.aspx, Accessed 20 September, 2013.

  • [2]

    H. L. Alder and E. B. Roessler, Introduction to probability and statistics, W.H. Freeman & Company, San Francisco, 1962.Google Scholar

  • [3]

    K. Alipour, S. A. A. Moosavian and Y. Bahramzadeh, Dynamics of wheeled mobile robots with flexible suspension: analytical modelling and verification, Int. J. Robot. Autom. 23 (2008), 242–250.Google Scholar

  • [4]

    C. H. Amon, S. Finger, D. P. Siewiorek and A. Smailagic, Integration of design education, research and practice at Carnegie Mellon University: a multi-disciplinary course in wearable computer design, in: Proceedings of Frontiers in Education Conference, IEEE 2 (1995), 14–22.Google Scholar

  • [5]

    C. M. Annappa and K. S. Panditrao, Application of value engineering for cost reduction – a case study of universal testing machine, Int. J. Adv. Eng. Technol. 4 (2012), 618–629.Google Scholar

  • [6]

    P. Aravinth, T. Muthu Kumar, A. Dakshinamoorthy and N. Arun Kumar, A criticality study by design failure mode and Effect analysis (FMEA) procedure in LINCOLN V350 PRO welding machine, Int. J. Adv. Eng. Technol. 4 (2012), 611–617.Google Scholar

  • [7]

    T. Baines, H. Lightfoot, G. M. Williams and R. Greenough, State-of-the-art in lean design engineering: a literature review on white collar lean, in: Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 220 (2006), 1539–1547.Google Scholar

  • [8]

    N. Bali, M. S. Boucherit and F. Boudjema, Influence of synthesis parameters of GPC and CDM: application to the PUMA manipulator, Int. Rev. Electr. Eng. 1 (2006), 410–417.Google Scholar

  • [9]

    B. Bao, Y. Yang, A. Liu, J. Zhao and L. Li, Task allocation optimization in collaborative customized product development based on adaptive genetic algorithm, J. Intell. Syst. 23 (2013), 1–19.Google Scholar

  • [10]

    M. R. Baxter, Product design: practical methods for the systematic development of new products. Chapman & Hall, London, 1995.Google Scholar

  • [11]

    R. Bernard, Early evaluation of product properties within the integrated product development. Shaker Verlag, Aachen, 1999.Google Scholar

  • [12]

    T. R. Browning, J. J. Deyst and S. D. Eppinger, Adding value in product development by creating information and reducing risk, IEEE Transact. Eng. Manage. 49 (2002), 443–458.Google Scholar

  • [13]

    T. R. Browning, E. Fricke and H. Negele, Key concepts in modelling product development processes, Syst. Eng. 9 (2005), 104–128.Google Scholar

  • [14]

    Cambridge University’s Engineering Department, Second year undergraduate integrated design project, Cambridge University Press, 2007.Google Scholar

  • [15]

    M. J. Chalupnik, C. M. Eckert, P. J. Clarkson, Modelling design processes to improve robustness, in: Proceedings of the 6th Integrated Product Development Workshop, IPD 2006, Schonebeck/Bad Salzelmen b. Magdeburg, Germany, 2006.Google Scholar

  • [16]

    M. J. Chalupnik, D. C. Wynn, C. M. Eckert and P. J. Clarkson, Understanding design process robustness: a modelling approach, in: Proceedings of the 16th International Conference on Engineering Design (ICED’07), Paris, France, pp. 455–456, 2007.Google Scholar

  • [17]

    H. I. Christensen, R. Dillmann, M. Hägele, A. Kazi and U. Norefors, European robotics, European Robotics Forum, July 2008.Google Scholar

  • [18]

    K. B. Clark and C. Y. Baldwin, Design rules, vol. 1: The power of modularity, MIT Press, Cambridge, MA, 2000.Google Scholar

  • [19]

    B. Clemson and D. Alasya, Implement TQM and CIM together, in: Proceedings of the International Engineering Management Conference, IEEE Engineering Management Society, New York (1992).Google Scholar

  • [20]

    J. Cohen, Statistical power analysis for the behavioural sciences, Academic Press, New York, 1988.Google Scholar

  • [21]

    M. A. Cohen, J. Eliashberg and T.H. Ho, An analysis of several new product performance metrics, J. Manuf. Serv. Operat. Manage. 2 (2000), 337–349.Google Scholar

  • [22]

    A. Coulibaly, R. Houssin and B. Mutel, Maintainability and safety indicators at design stage for mechanical products, Comput. Ind. 59 (2008), 438–449.CrossrefWeb of ScienceGoogle Scholar

  • [23]

    A. W. Court, Issues for integrating knowledge in new product development: reflections from an empirical study, Knowl-Based Syst. 11 (1998), 391–398.Google Scholar

  • [24]

    J. Eschenbaecher and F. Graser, Managing and optimizing innovation processes in collaborative and value creating networks, Int. J. Innov. Technol. Manage. 8 (2011), 373–391.CrossrefGoogle Scholar

  • [25]

    Fanuc Robotics, M410 ib series, Product Guide. Available at: http://www.fanuc.co.jp/en/product/catalog/pdf/M-410iB(E)_v07_s.pdf. Accessed 20 September, 2013.

  • [26]

    C. Huang, Overview of modular product development, Proc. Natl. Sci. Council, ROC(A) 24 (2000), 149–165.Google Scholar

  • [27]

    F. Khalaf and K. Yang, Product development processes – from deterministic to probabilistic: a design for 6-sigma approach to lean product validation, part II, Int. J. Prod. Dev. 3 (2006), 18–36.Google Scholar

  • [28]

    V. Krishnan, Modeling ordered decision making in product development, Eur. J. Oper. Res. 111 (1998), 351–368.CrossrefGoogle Scholar

  • [29]

    V. Lévárdy and T. R. Browning, Adaptive test process – designing a project plan that adapts to the state of a project, INCOSE Publications, 2005.Google Scholar

  • [30]

    D. Liu, R. Deters and W. J. Zhang, Architectural design for resilience, Enterprise Information Systems (August 2009), 1–16.Google Scholar

  • [31]

    R. Maul and D. Tranfield, Methodological approaches to the regeneration of competitiveness in manufacturing, in: Proceedings of 3rd International Conference on Factory 2000, IEE, UK, pp. 12–17, 1992.Google Scholar

  • [32]

    S. A. A. Moosavian and K. Alipour, On the dynamic tip-over stability of wheeled mobile manipulators, Int. J. Robot. Autom. 22 (2007), 322–328.Web of ScienceGoogle Scholar

  • [33]

    K. N. Otto and E. K. Antonsson, Trade-off strategies in engineering design, Res. Eng. Des. 3 (1991), 87–104.CrossrefGoogle Scholar

  • [34]

    G. Pahl and W. Beitz, Engineering design: a systematic approach, Springer-Verlag, New York, 1998.Google Scholar

  • [35]

    R. D. Pearce, Decentralised R&D and strategic competitiveness: globalised approaches to generation and use of technology in multinational enterprises MNEs, Res. Policy 28 (1999), 157–178.CrossrefGoogle Scholar

  • [36]

    D. N. Prabhakar Murthy, M. Rausand and T. Østerås, Product reliability: specification and performance, Springer, London, 2008.Google Scholar

  • [37]

    U. Sekaran, Research methods for business, John Wiley & Sons Inc., New York, 2003.Google Scholar

  • [38]

    C. Stummer and R. Vetschera, Decentralized planning for multiobjective resource allocation and project selection, Cent. Eur. J. Oper. Res. 11 (2003), 253–279.Google Scholar

  • [39]

    D. D. Wackerly, W. Mendenhall and R. L. Scheaffer, Mathematical statistics with applications, Duxbury Press, New York, 1996.Google Scholar

  • [40]

    J. P. Womack and D.T. Jones, Lean thinking, Simon and Schuster, New York, 1996.Google Scholar

  • [41]

    J. Womack, D. Jones and D. Roos, The machine that changed the world, Macmillan, New York, 1990.Google Scholar

  • [42]

    H. Yavuz, An integrated approach to the conceptual design and development of an intelligent autonomous mobile robot, Robot. Auton. Syst. 55 (2007), 498–512.CrossrefWeb of ScienceGoogle Scholar

  • [43]

    Y. Y. Yusuf, M. Sarhadi and A. Gunasekaran, Agile manufacturing: the drivers, concepts and attributes, Int. J. Prod. Econ. 62 (1999), 33–43.CrossrefGoogle Scholar

About the article

Salah A.M. Elmoselhy

Salah A. M. Elmoselhy holds an MS in mechanical design and production engineering that he received from Cairo University. He also holds an MBA in international manufacturing business that he received from Maastricht School of Management (MSM), Maastricht University. He has 10 years of industrial experience in CAD/CAM and robotized manufacturing systems. He was recently a researcher at the Engineering Department and Fitzwilliam College of Cambridge University from which he received a diploma of postgraduate studies in engineering design. He is currently a PhD candidate in mechanical engineering, working with the International Islamic University Malaysia (IIUM) and the Center for Sustainable Mobility at Virginia Polytechnic Institute and State University (Virginia Tech). He has authored/coauthored about 20 refereed publications in journals, including ISI-indexed journals and SCOPUS-indexed journals, and international conferences. His research appears in esteemed journals such as the Journal of Manufacturing Systems, SAE Transactions: Journal of Materials and Manufacturing, and International Journal of Vehicle Systems Modelling and Testing. His research interests include operations management, innovation management, engineering design, robotics, modeling, and simulation and system dynamics. He is an associate member of the IMechE.


Corresponding author: Salah A.M. Elmoselhy, Associate Member of IMechE and Former Researcher at Fitzwilliam College, Cambridge, UK, e-mail:


Received: 2014-02-22

Published Online: 2014-08-20

Published in Print: 2015-03-01


Citation Information: Journal of Intelligent Systems, Volume 24, Issue 1, Pages 117–134, ISSN (Online) 2191-026X, ISSN (Print) 0334-1860, DOI: https://doi.org/10.1515/jisys-2014-0024.

Export Citation

©2015 by De Gruyter.Get Permission

Comments (0)

Please log in or register to comment.
Log in