ARTICLE IN PRESS

Applied Ergonomics 39 (2008) 316–324 www.elsevier.com/locate/apergo

Development and design of a dynamic armrest for hydraulic-actuation joystick controlled mobile machines T. Murphy, M.L. Oliver School of Engineering, University of Guelph, Ont., Canada N1G 2W1 Received 17 May 2007; accepted 24 September 2007

Abstract Standard armrests used in conjunction with joysticks of heavy mobile machinery have been proven to inadequately meet operator needs, resulting in excessive static loading of shoulder musculature. During joystick operation, the trajectory of the user’s forearm is governed by the motion of the controller, which creates horizontal and vertical movement of the forearm. The vertical motion of the forearm in the forward and backward motion create postures that stationary armrests cannot support thereby generating increased muscle activation and risk of repetitive strain injuries. The current paper describes the design process used in creating a dynamic armrest that replicates the operator’s natural motion trajectories. By incorporating the natural motion paths into a dynamic armrest, the postural requirements and muscular activation of the operator’s shoulder may be reduced. r 2007 Elsevier Ltd. All rights reserved. Keywords: Armrest; Joystick; Ergonomics

1. Introduction The operation of heavy mobile machines often involves long and highly fatiguing shifts of repetitive joystick movements. Past research has indicated that during joystick manipulation, muscle activation in the shoulder is at, or above, a constant static load of 2% maximum isometric voluntary contraction (MVC) (Asikainen and Harstela, 1993; Attebrant et al., 1997; Lindbeck, 1982; Nakata et al., 1993). Static muscular loads exceeding 2% MVC have been considered by many to be inappropriate and can lead to the development of repetitive strain injuries (Jonsson, 1978). Others feel that a more acceptable upper limit is 1% MVC unless adequate rest periods are given (Aara˚s and Westgaard, 1987). Research by Lindbeck (1985) assessed shoulder muscle activation (upper trapezius (UT), anterior deltoid (AD), middle deltoid, and posterior deltoid (PD)) while using joysticks of different stiffnesses and ranges of motion. Corresponding author. Tel.: +1 519 824 4120x52117; fax: +1 519 836 0227. E-mail address: [email protected] (T. Murphy).

0003-6870/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.apergo.2007.09.002

Conclusions indicate that higher joystick resistances and movement spans increase muscle activity in the middle and PD and to a lesser extent the AD and the UT. A similar study by Northey (2004) established opposing results, stating that the activation in the PD was not affected by joystick stiffness. Both of the previously mentioned studies do, however, agree that the UT is not affected by joystick stiffness. This indicates that the UT is almost entirely dedicated to stabilizing the shoulder complex during joystick manipulation. It should also be noted that results between the two studies may differ due to slightly different joystick designs. Typical North American joysticks are hydraulic and include balance and return springs which require increased operator input forces when compared to the Scandinavian hand levers that are typically electronic requiring smaller input forces (Oliver et al., 2000, 2006). Northey (2004) also found that the shoulder muscles (UT, PD and AD) were most active during forward motions of operation, with peak amplitudes of the enveloped EMG signals were of 71%, 43%, and 39% of task-related MVC, respectively. In all directions (forward, backward, outward and inward), the UT had the highest

ARTICLE IN PRESS T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

peak EMG values except in the backward direction where only the PD was higher. A study by Attebrant et al. (1997) compared shoulder musculature loading for two types of joysticks and a horizontally moving armrest. The first type of joystick was a pronated hand lever and the second type was a minilever. Previous to this study, operators using pronated hand levers were shown to have increased symptoms of pain in both the elbow and the shoulder region, requiring more sick leave compared to operators using a neutral joystick grip (Gellerstadt, 1997; Grevsten and Sjo¨gren, 1996; Hagberg and Liden, 1991). Particularly during inward motion, the pronated position places the most strain in the shoulder due to increased elbow elevation (or shoulder abduction). As a consequence, shoulder rotation is increased as the hand is further pronated, therefore further activating shoulder musculature (Grevsten and Sjo¨gren, 1996). The second type of joystick studied, the mini-lever, was shown to reduce UT muscle activity as a result of decreasing the range of shoulder motion. Similar results were found by Lindbeck (1986). By decreasing the input range of the lever, the movement span of the operator’s hand and shoulder are also decreased thus allowing the arm to remain supported by the armrest throughout the range of motion. The external support consequently leads to reduced muscle activation in the UT by reducing its role as a stabilizer (Asikainen and Harstela, 1993; Roman-Liu et al., 2001). Although studies have shown that mini-levers slightly decrease shoulder muscle activation (and pain prevalence), they were not shown to reduce the prevalence of pain in the neck (Grevsten and Sjo¨gren, 1996). Minilever operators often complain of difficulties in activating electrical switches while concomitantly using mini-levers (Hagberg and Linden, 1991) helping to explain the mixed results. A study of neutral-hand-levers compared muscle activation in the UT and deltoid between work with and without arm supports (Hansson, 1990). While the authors found no significant differences between the two experimental conditions, it should be noted that three quarters of the subjects experienced decreased muscle activity with the armrest except when the lever was pulled in reverse. As alluded to by Attebrant et al. (1997), this result is partially explained by the restrictive design associated with the stationary armrests. Alternatively, the load on the UT may be reduced through the introduction of a more effective armrest design, which was also attempted during the aforementioned study by Attebrant et al. (1997). Although armrests have generally been considered to reduce muscle activation they continue to fail at providing adequate support across the entire movement range (Attebrant et al., 1997; Northey, 2004). During forward controller motion, the amplitude of the arm movement requires the forearm to be raised from the armrest, requiring the UT to provide the majority of the stabilization (Northey, 2004). During backward

317

joystick motion, the armrest resists natural arm pendulation, necessitating shoulder elevation (Lindbeck, 1982; Northey, 2004) also requiring increased UT activation through increased stabilization. Although Attebrant et al. (1997) concluded that both mini-levers and horizontally moving arm supports reduced UT loads, the reduction was small. The seemingly small reduction in muscular loading could be a result of the experimental conditions used. The study featured a non-pronated, shorter lever (98 mm) that was extended from the armrest. By comparison, typical North American controllers are often semipronated with a length of 240 mm and require movements of larger amplitudes. Consequently, larger movement amplitudes increase the vertical movements of the forearm that are produced as the arm pendulates over the joystick. Small movements are accompanied by less vertical forearm movement and therefore more contact with the armrest, as well as lower UT muscular loading. Therefore, it is conceivable to assume that the decrease in UT loading may have been more drastic if the analysis were to have compared the mini-lever to a controller with a longer lever length. Since the present study is concerned with North American joysticks, the vertical movement of the armrest is of greater importance. The implementation of a pendulating armrest, one that will move with two degrees of freedom in the sagittal plane (translation in both vertical and horizontal directions), will allow for a more support throughout the movement cycle. An armrest that is capable of following the natural movement of the forearm should provide support throughout the range of motion therefore requiring less stabilization from the shoulder, and more specifically, lower UT activation. The three main objectives of this paper are to: (i) determine if an arm supported throughout joystick motion actually decreases the activation levels of the UT, (ii) determine the unencumbered natural motion path of an operator’s arm while operating hydraulic-actuation controller in the forward and backward directions, and (iii) implement a simple mechanism that mimics the natural motion path to ergonomically support the arm throughout the movement range of joystick manipulation. 2. Methods Subjects were seated in laboratory mock-up of an excavator cab (Fig. 1). The setup resembled the geometry of a common North American Excavator cab fitted with a right-handed joystick mounted onto a frame that supported the seat. Armrests were removed so that the operator’s arm motion was not affected by the armrest. Subjects were then fitted to the seat to keep their feet flat on the floor with their knees at an ergonomically accepted neutral position of 901 flexion. Subjects were able to sit comfortably upright in the chair with the back support reclined 151 from vertical. The joystick was also adjustable

ARTICLE IN PRESS 318

T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

Fig. 2. Joystick and holding device used on cab mock-up frame. Arrows indicate the direction each adjustment knob moves the holding device.

Fig. 1. Laboratory mock-up of a typical North American excavator chair and right-handed joystick. The coordinate system is arranged with the xaxis directed anteriorly to the subject sitting in the chair.

to account for subject stature (Fig. 2), and for each subject it was adjusted so that when the subject gripped the joystick, the forearm was horizontal and the wrist in a neutral posture. Once subjects were seated in the neutral posture, a four-point harness was secured over each shoulder and across the lap (Fig. 3). The harness was used to stabilize the shoulder and restrict the subject’s arm movement to exclude any trunk movement. 2.1. Preliminary testing—proof of concept The rationale for the redesign of the current armrest was to reduce muscular activity of the shoulder muscular of the operators. Therefore, in order to avoid unnecessary design costs, a small pilot study was undertaken to assess the prospect that a fully supported arm would create a reduction in shoulder muscle activation during joystick manipulation. The pilot study involved a single subject manipulating the joystick in forward and backward directions during three support conditions: no arm support, moderate arm

Fig. 3. Operator sitting in a mock-up of a typical North American excavator cab fitted with the standard armrest. Trunk movement was minimized through the use of a four-point harness.

ARTICLE IN PRESS T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

support and full arm support. Arm supports were simply removed for the no arm support condition while moderate arm support utilized the geometry of a standard arm support found in a typical North American excavator. The expertise of a certified physiotherapist was called upon to provide full support by gently supporting and not guiding the user’s arm. Five cycles of each direction (forward and backward) were performed for each support condition. One joystick cycle consisted of joystick motion from the neutral position, to the end of range of joystick, and returning to the neutral position. EMG signals from the UT, PD, and AD were collected and analyzed by full wave rectifying the signal and then linear enveloping using a second order, 6 Hz, dual pass Butterworth filter (Winter, 1990). Peak and mean EMG values were then determined from the linear enveloped EMG data and compared among the three armrest conditions.

2.2. Subjects Five right-handed male subjects, with no previous experience of using hydraulic-actuation joysticks, between 20 and 26 years of age (mean7S.D.: age 2472.7 years; height 17672.6 cm, mass 6574.8 kg) completed the study. Subjects had no history of musculoskeletal disorders or shoulder pain and were familiarized with the setup and equipment prior to testing. Since this was a proof of concept design, subjects were selected by height to represent the 50th percentile male population (Kroemer et al., 1994). Approval from the University of Guelph Ethics Committee to conduct the study was obtained prior to testing, and each subject provided informed consent.

2.3. Test protocol Subjects performed three repetitions of each forward and backward joystick movements without visual feedback of the controller in a randomized order while seated in the neutral posture. At the time of motion capture the researcher provided verbal feedback to the subject to ensure that the joystick motion remained on axis. Trials which were not visually acceptable were then repeated. All subjects gripped the joystick in the same manner at a specified position as marked on the handle of the joystick. The tasks involved one motion per trial beginning at the neutral position with the directional command ‘‘forward’’ or ‘‘backward’’ and finished once the joystick was returned to the neutral position. Subjects were not given any indication of speed at which to execute the motion, but only to attempt to minimize off-axis motion and maintain continuous motion within the trial.

319

2.4. Kinematic analysis Kinematic data were captured by six Vicons M2 cameras (Vicon Peak, Oxford, UK) with a frame rate of 100 Hz using a total of nine retro-reflective markers. Markers were placed at the acromio-clavicular joint of the right shoulder, the lateral epicondyle of the elbow, the styloid process of right ulna, as well as six arranged on the joystick to capture its motion. Trajectories of wrist and elbow marker were then used to represent the forward and backward (x-axis) and vertical motion (z-axis) of the forearm (see Fig. 1 for reference to coordinate system). Lateral motion of the arm was ignored since preliminary tested showed that there was negligible (o15 mm) lateral motion (i.e., motion along the y-axis). 2.5. Data processing Raw data were processed using MatlabTM (Version 6.5; The Mathworks Inc., 2001, Natick, MA). The kinematic data were first clipped using the first derivative of joystick angle, leaving only the data describing the respective forward or backward motion. Next, all kinematic data were re-sampled to create trials of a length that represented the average motion cycle time. Trials with excessive lateral elbow deviation (415 mm) resulting from shoulder abduction were excluded from the data set to ensure that the forearm motion was accurately represented in the x and zaxes. A value of 15 mm was selected based on the 50th percentile male’s arm length to produce less than 31 of shoulder abduction. This approximation was based on a simple rigid-bodied two-link (upper and lower arm) algebraic model that assumed the scapula was stationary and the shoulder behaved as a ball and socket joint in flexion (Veeger et al., 1997). All trials, for all subjects, meeting the previous criteria were then averaged to create an average motion for each direction, resulting in trajectories that represented the ‘‘natural motion path’’ of the operator’s forearm. The natural motion path trajectories of the wrist and elbow were then integrated in the design of the armrest to create a dynamic armrest with identical trajectories. Since the armrest travels under the arm, the trajectories were required to be offset in the z-axis to preserve the natural motion path of the forearm. The wrist trajectory was also shifted posteriorly to account for the close proximity to the joystick to the wrist. This was necessary to allow the armrest to move freely with out being obstructed by the structural aspects of the joystick. The modified trajectories are referred to as ‘‘corrected trajectories’’. The corrected trajectories of the wrist and elbow were fitted accurately (R40.95) with third-order polynomials to create smoother armrest trajectories. The polynomials were extended in both the forward and backward directions to allow for a small amount of extra movement in the armrest to ensure that the armrest would provide unrestricted motion. The smoothed trajectories were then integrated to

ARTICLE IN PRESS T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

320

above), which were then incorporated into the dynamic armrest’s design to control the trajectory of the armrest. The smoothed armrest trajectories are illustrated in Figs. 6 and 7. Once the smoothed trajectories were obtained, they were then used to create dedicated bearing races which defined the motion path of the dynamic armrest (Patent Pending). Each of the four bearing races were milled from individual support members made of 6061 aluminum. Support members were then fitted to a support structure constructed of 1018 mild steel that was then secured to the frame of the excavator chair. Four, 19 mm deep groove ball bearings (NSK Ltd., Tokyo, Japan) were used to create near frictionless translation of the armrest within the bearing race of each support. Front and rear bearing pairs were connected via axels which were fixed to the armrest. The complete design is illustrated in the isometric form in Fig. 8 (Patent Pending). Fig. 9 demonstrates the armrest in various positions throughout the entire range of motion of the joystick.

create the armrest path of the newly designed dynamic armrest. 3. Results Results of the proof-of-concept testing showed that for all muscles, the fully supported arm displayed the lowest EMG variables during both forward and backward directions. Meanwhile, the moderate support and the no support conditions illustrated similar EMG values for the forward direction, while the moderate support displayed the highest EMG variables for forward motion. The former result further highlights the apparent deficiencies of current armrests, namely that they increase shoulder activity by hindering natural arm motion. A full summary of the EMG variables can be seen in Fig. 4. Although the pilot work lacks statistical power (n ¼ 1), it clearly demonstrates the need for further investigation of dynamic armrests and current armrests used by joystick operators. Based on these findings, it was deemed appropriate to carry on with the development of a dynamic armrest that mimicked the forearm trajectories of joystick users. The natural motion path of the user’s wrist and elbow are illustrated in Fig. 5. Trajectories are referenced relative to the elbow marker when the joystick is pulled in the backward position. The wrist and elbow trajectories were then used to create the corrected trajectories (as described

4. Discussion It is essential to remember that the current dynamic armrest is a proof-of-concept design and intended solely for the laboratory setting. The objective of the design was to confirm that an armrest that replicated the natural motion patterns of the user’s forearm, and thereby provide

Peak muscle activation during forward joystick motion

Peak muscle activation during backward joystick motion

0.03

0.04

0.025

0.035

EMG (uV)

EMG (uV)

0.03 0.02 0.015 0.01

0.025 0.02 0.015 0.01

0.005

0.005

0.0 1

0.0 2

1

3

Mean muscle activation during forward joystick motion

3

Mean muscle activation during backward joystick motion 0.02

0.018 0.016

0.018

0.014

0.016 0.014 0.012

0.012

EMG (uV)

EMG (uV)

2

0.01 0.008 0.006

0.01 0.008 0.006 0.004 0.002

0.004 0.002

0

0 1

2

1

3

Full support

Moderate support

2

3

No support

Fig. 4. Summary of peak and mean EMG values for fully supported, moderately supported and unsupported arm during forward and backward joystick manipulation.

ARTICLE IN PRESS T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

321

40

z-axis: vertical movement (mm)

35 30 25 20 15 10 5 0 -50

-5

0

50

100

150

200

250

300

350

x-axis: horizontal movement (mm)

Wrist Trajectory

Elbow Trajectory

Fig. 5. Corrected motion path of elbow and wrist during unrestricted joystick manipulation with displacement relative to elbow position when the controller is pulled to the full extent of the backward direction. Not to scale.

Z Displacement in the vertical direction (mm)

35 30 25 y = 0.00x 3 + 0.00x2 + 0.32x + 1.63 20 0.9969 R2 = 1.00 15 10 5 0

-75

-55

-35

-15

-5

5

25

45

65

-10 -15 X Displacement in the forward/backward direction (mm) Elbow movement

Armrest movement

Z Displacement in the vertical direction (mm)

Fig. 6. Average motion path of joystick operator’s elbow as collected from motion analysis, then fitted with a third-order polynomial representing the trajectory of the dynamic armrest under the operator’s wrist.

-70

1.5 1 0.5 -50

-30

0 -10 -0.5

10

30

50

70

-1 -1.5 -2 -2.5

y = 0.00x 3 - 0.00x 2 + 0.01x + 0.66 R2 = 0.93

-3 -3.5 -4 X Displacement in the forward /backward direction (mm) Wrist movement

Armrest movement

Fig. 7. Average motion path of joystick operator’s wrist as collected from motion analysis, then fitted with a third-order polynomial representing trajectory of the dynamic armrest under the operator’s elbow.

ARTICLE IN PRESS T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

322

support throughout the range natural motion, would reduce muscular activation in the shoulder. A companion paper reports on the experiment which confirms this (Murphy and Oliver, submitted for publication). Since the forward and backward directions were shown to be the most strenuous for the UT (Northey, 2004) these were the motions tackled first. Lateral motions were excluded to ensure that the concept of the dynamic armrest was tested and not the functionality of the design. Similarly, the current device was designed to fit the chair in the laboratory setting and exceeds size restrictions that would limit its feasibility in working heavy mobile machinery. As previously mentioned, the stature of the five subjects (176.370.3 cm) that participated in the data collection closely represented the 50th percentile male (175.6 cm) as described by Kroemer et al. (1994). Consequently, the design should most closely fit subjects of the following characteristics: stature of 175.6 cm and humeral length of 36.7 cm (Kroemer et al., 1994). However, the change in humeral length from 5th percentile to 50th percentile or 50th percentile to 95th percentile is less than 4.2 mm; assuming that the shoulder acts as a pivot, the radius of curvature that the elbow travels will then differ by less than 4.2 mm from 50th percentile to either stature extreme. Consequently, the design should accommodate the ergonomically accepted range of 90% of all male users (Bridger, 2003). This calculation assumes that the shoulder rotates forward motion of armrest

forward motion of armrest

forward motion of armrest

Fig. 8. Dynamic armrest design for right-handed hydraulic-actuation joysticks in a laboratory setting (Patent Pending).

about a fixed axis, and although this is not completely accurate, it is a close approximation based on a stationary shoulder (Nordin and Frankel, 1989) and is a reasonable approximation given the relatively small range of upper extremity motion during joystick manipulation. In addition, the armrest is adjustable in the vertical direction to allow for various heights thus allowing for various heights of operators. Research regarding the benefits of armrests has been conflicting with some authors reporting them to be effective in reducing UT load (Westgaard and Aaras, 1985), while other reports fail to demonstrate any changes in loading (Hedge and Powers, 1995). In many cases, the disadvantage to arm supports can often be their usefulness over a large workspace. In situations where arms are required to cover large workspaces, armrests often cannot provide adequate support throughout the entire movement range of the task. The success of an armrest support is often dependant upon the task, as well as the efficiency of the armrest to provide the required support over the entire range of motion. In many cases where forearm support may be required, the challenge of implementing a design which is supportive as well as unrestricting can be difficult. The lack of efficiency over the entire workspace may at least partially explain why findings regarding the benefits of arm supports are equivocal. Given that arm movement is very constrained and repeatable during controller operation, the ability of the armrest to efficiently cover the workspace is less of a concern. Odell et al. (2007) designed and tested a dynamic arm support that was capable of spanning the work envelope for a large number of common hand intensive tasks. Results demonstrated that the device could reduce the static muscular loads in some of the upper extremity muscles. Although the Odell et al. (2007) design and current design are both dynamic supports, the design functionalities differ. The movement of the current device is dictated by the exact forearm trajectories associated with joystick manipulation, whereas the motion of the Odell et al. (2007) design is directed by the user’s motion. Both designs likely have their advantages and disadvantages but it is suspected that by reducing the internal inertia and friction levels of the support, a reduction in muscle activation level requirements may be achieved. One objective of this paper was to determine the natural motion path of an operator’s arm while manipulating a

Fig. 9. Demonstration of dynamic armrest in backward, neutral, and forward positions of joystick operation.

ARTICLE IN PRESS T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

typical North American hydraulic-actuation controller in the forward and backward directions. Another objective was to implement this natural motion path into a newly designed, ergonomically correct arm support that mimics the user’s motion during joystick manipulation. The success has been established partly on the armrest’s ability to decrease muscle activation in the UT, PD, and AD during forward and backward motions. The subjective response of the user was another critical evaluation tool used to assess the efficacy of the dynamic armrest. These aspects are reported on in a companion paper by Murphy and Oliver (submitted for publication). Although concept and prototype suggest that a reduction in muscle activation in the shoulder can be achieved by fully supporting the user’s forearm the utility of the device must also be proven by end user. Two important modifications that must be addressed during subsequent design iterations including assessing the need for rotational movement of the armrest during lateral controller operation, and incorporating retrofitting capabilities. These aspects were ignored during this proof-of-concept design to ensure that the results of future testing were complication associated with the functionality of the design. It is possible, based on the success of this design, that this novel ergonomic approach of implementing a dynamic arm support can be applied to other joystick applications including machinery in the forestry, construction, and mining industries. It is also possible that this approach can transcend tasks involving joysticks to include any task that is of a constrained and repetitive nature, including tasks such as assembly lines. 5. Conclusions A novel approach has been presented for the design of an arm support that can be used in constrained and repetitive work environments. More specifically, the development of a dynamic armrest has been described that mimics the unencumbered forearm motion of excavator operators using a semi-pronated, hydraulic-actuation joystick. By accurately measuring the unencumbered forearm trajectories of five, 50th percentile male operator’s manipulating a joystick from an excavator their forearm motions were successfully captured and averaged. The averaged trajectories of the wrist and forearm were used to design a simple arm support mechanism which mimicked the operator’s movement during forward and backward joystick use. A thorough investigation regarding the efficacy of the newly design dynamic armrest is discussed in a companion paper (Murphy and Oliver, submitted for publication). Acknowledgments The authors gratefully acknowledge the financial support provided by the Natural Sciences and Engineering

323

Research Council of Canada, Canadian Foundation for Innovation, and the Ontario Innovation Trust. References Aara˚s, A., Westgaard, R.H., 1987. Further studies of postural load and musculoskeletal injuries of workers at an electo-mechanical assembly plant. Appl. Ergon. 18, 212–219. Asikainen, A., Harstela, P., 1993. Influence of small control levers of grapple loader on muscle strain, productivity and control errors. Int. J. Forest Eng. 5, 23–28. Attebrant, M., Winkel, J., Mathiassen, S.E., Kjellberg, A., 1997. Shoulder–arm muscle load and performance during control operation in forestry machines: effects of changing to a new arm rest; lever and boom control system. Appl. Ergon. 28, 85–97. Bridger, R.S., 2003. Introduction to Ergonomics, second ed. McGrawHill, New York. Gellerstadt, S., 1997. Mechanised cleaning of young forest—the strain on the operator. Int. J. Ind. Ergon. 20, 137–143. Grevsten, S., Sjo¨gren, B., 1996. Symptoms and sickleave among forestry machine operators working with pronated hands. Appl. Ergon. 27, 277–280. Hagberg, J.S., Lide´n, E., 1991. Mini-controls decrease musculoskeletal symptoms’ (Minispakar minskar belastningsbesva¨r) Forskningsstiftelsen Skogsarbetaren (Box 1184, S-164 22 Krista). Resultat nr 24 (in Swedish), p. 4. Hansson, J., 1990. Ergonomic design of large forestry machines. Int. J. Ind. Ergon. 5, 255–266. Hedge, A., Powers, J.R., 1995. Wrist postures while keyboarding: effects of a negative slope keyboard system and full motion forearm supports. Ergonomics 38, 508–517. Jonsson, B., 1978. Kinesiology—with special reference to electromyographic kinesiology. In: Cobb, W.A., Van Duijn, H. (Eds.), Contemporary Clinical Neurophysiology (EEG Suppl. #34). Elsevier Scientific Publishing Co., Amsterdam. Kroemer, K.H.E., Kroemer, H.B., Kroemer-Elbert, K.E., 1994. Ergonomics: How to Design for Ease and Efficiency. Prentice-Hall, Englewood Cliffs. Lindbeck, L., 1982. Armsto¨dets betydelse fo¨r avlastning av skuldran vid spakmano¨vrering. Underso¨kningsrapport. 1982: 35. Arbetarskyddsstryrelsen. Lindbeck, L., 1985. Skulderbelastning vid reglagearbete med tre olika spakmotsta˚nd. Underso¨kningsrapport. 1985: 12. Arbetarskyddsstyrelsen. Lindbeck, L., 1986. Na˚gra effekter av minskat spakutslag vid reglagearbete. Underso¨kningsrapport. 1986: 15. Arbetarskyddsstyrelsen. Murphy, T., Oliver, M., (submitted for publication). Dynamic armrest for hydraulic-actuation controller use. Appl. Ergon. Nakata, M., Hagner, I., Jonsson, B., 1993. Trapezius muscle pressure pain threshold and strain in the neck and shoulder regions during repetitive light work. Scand. J. Rehabil. Med. 25, 131–137. Nordin, M., Frankel, V.H., 1989. Basic Biomechanics of the Musculoskeletal System, second ed. Lea & Febiger, Philadelphia. Northey, G.W., 2004. The effects of stiffness and speed on upper limb electromyography during joystick use. Masters Thesis (M.Sc. (Eng)), School of Engineering, University of Guelph, Ont., Guelph. Odell, D., Barr, A., Goldberg, R., Chung, J., Rempel, D., 2007. Evaluation of a dynamic arm support for seated and standing tasks: a laboratory study of electromyography and subjective feedback. Ergonomics 50, 520–535. Oliver, M., Rickards, J., Biden, E., 2000. Off-road machine controls: investigating the risk of carpal tunnel syndrome. Ergonomics 43, 1887–1903. Oliver, M., Rogers, R., Rickards, J., Tingley, M., Biden, E., 2006. Effect of stiffness and speed on selected dynamic torque characteristics of hydraulic-actuation joystick controls for heavy vehicles. Ergonomics 49, 249–268.

ARTICLE IN PRESS 324

T. Murphy, M.L. Oliver / Applied Ergonomics 39 (2008) 316–324

Roman-Liu, D., Tokarski, T., Kaminska, J., 2001. Assessment of the musculoskeletal load of the trapezius and deltoid muscles during hand activity. Int. J. Occup. Saf. Ergon. 7, 179–193. Veeger, H.E.J., Yu, B., An, K.-N., Rozendal, R.H., 1997. Parameters for modeling the upper extremity. J. Biomech. 30, 647–652.

Westgaard, R.H., Aaras, A., 1985. The effect of improved workplace design on the development of work-related musculo-skeletal illnesses. Appl. Ergon. 16, 91–97. Winter, D.A., 1990. Biomechanics and Motor Control of Human Movement, third ed. Wiley, Toronto.