TECHNOLOGY AND SPORT

Vicon assists in unraveling illegal bowling dilemma

by Jacque Alderson, Bruce Elliott, Siobhán Reid, Marc Portus and David Lloyd

Down Syndrome Research Foundation and Centre for Human Movement Analysis

Queen Alexandra Centre for Children's Health, Victoria, B.C., Canada

Technology is playing an ever-increasing role in today's society. However, the use of technology to assess the legality of a bowling action in cricket is a more recent occurrence. With players now competing for very large contracts, it is imperative that technique assessment produces similar results from different laboratories and over repeated trials from the same laboratory. Where possible, results must also accurately reflect what happens under match conditions. The School of Human Movement and Exercise Science (HM&ES) recently acquired a 12 camera, Vicon 612 System and have used it to assess the legality of the action of cricket spin and fast bowlers for National Cricket Boards and the International Cricket Council (ICC).

The “throwing” Rule in Cricket

For those not familiar with the game of cricket or the bowling action, law 24.3 in The Laws of Cricket (2000 Code 2nd Edition-2003) provides the following definition of a fair delivery:

“A ball is fairly delivered in respect of the arm if, once the bowler's arm has reached the level of the shoulder in the delivery swing, the elbow joint is not straightened partially or completely from that point until the ball has left the hand. This definition shall not debar a bowler from flexing or rotating the wrist in the delivery swing.”

At the recommendation of the biomechanics community, cricketing authorities have recently conceded that all bowlers will display some degree of passive elbow extension due to the inertial load encountered during the bowling action. This concession has resulted in the ICC setting elbow extension ‘tolerance levels' of 10º for fast bowlers, 7.5 º for medium pace bowlers and 5 º for spin bowlers. It is interesting to note that fast bowling research by Portus and colleagues (2001), which studied thirty-four deliveries bowled by twenty-one first class fast bowlers, reported 14 in excess of the 10º limit. Irrespective of what motion analysis system is used to collect data, the accuracy of kinematic data is a combination of firstly, the “biomechanical model” used to define joint centers and joint axes of rotation, and secondly, the environment in which the data are captured.

In traditional joint centered models where markers are positioned over selected anatomical landmarks to define joint centers (eg, wrist, elbow and shoulder), there are significant errors in joint center and axes of rotation definitions due to excessive skin movement. This is particularly evident in full elbow extension where external markers placed on elbow epicondyles will result in elbow flexion artefact. This error can be attributed to the difficulty in affixing markers on the epicondyles during full elbow extension, such that the markers accurately represent the elbow's transepicondylar (TE) axis (assumed flexion/extension axis). The problem is further exacerbated in situations where individuals have joint abnormalities and where the TE axis is not the best representation of the flexion/extension axis of the elbow. In much the same way as gait analysis data can be affected by ‘cross-talk', particularly in individuals with disabilities (e.g. cerebral palsy, spina bifida), the same problem occurs in the upper limb. This is further compounded when three dimensional angles between the shoulder, elbow and wrist are projected onto a two dimensional plane for data interpretation and reporting purposes. It is also interesting to note that the two most reported bowlers in recent years, Muttiah Muralitharan and Shoaib Akhtar (Figure 1) have significant abnormalities at the elbow joint.

Figure 1: Shoaib Akhtar displaying upper arm external rotation and elbow hyperextension.

The second point here relates to the environment in which the data are captured. In the ideal situation data are collected during a match. However in our considered opinion, the compounding error of recording footage from a distance (boundary) digitizing joint centers through clothing, and the modelling error associated with a traditional joint centered approach would exceed the tolerance levels set by the ICC, particularly with reference to the 5º limit allocated to spin bowlers. Research is about to commence at the School of HM&ES to determine the measurement error in bowling arm motion reconstructed from match versus laboratory conditions.

Figure 2: Three dimensional reconstruction of Vicon markers and the shoulder joint complex from CT scans.

For the reasons previously mentioned and in light of the stringent tolerance levels set by the ICC, the biomechanics group at UWA determined that the only accurate way to measure the motion of the bowling arm was to develop a ‘fully' three dimensional biomechanical model of the upper body. The group also believe that for accuracy, testing should occur in a laboratory environment using an optical motion analysis system shown to be one of the most accurate commercial motion analysis systems available (Richards, 1999). Over a period of five years the biomechanics group has been developing and validating a “Full Body Model” for the analysis of movement (Lloyd et al., 2000; Besier et al., 2003). These customized models have been written using Vicon Bodybuilder BodyLanguage software. What follows is a general description of the “Upper Body Model” related to the bowling arm in cricket and the data collection procedures employed in a recent biomechanical analysis of Muttiah Muralitharan, conducted on behalf of the Sri Lankan Cricket Board and the ICC.

Figure 3: Upper Body Model, not all markers can be seen.

(Picture courtesy of Tom Rovis-Hermann, West Australian newspaper)

An upper arm triad, which can best be seen on both upper arms in Figure 3, is placed such that two markers run parallel with the long axis of the humerus, while the third marker is oriented medially. Imagine the difficulty in trying to digitize the shoulder center from the views obtained from three video cameras, when the bowler is wearing a shirt during match conditions. While it may be relatively easy to identify the joint center from one direction, the ability to also identify this same point from other camera views when obscured by clothing and the body itself, makes this task extremely difficult.

Elbow Center and Forearm Motion

Determination of the elbow joint center is critical to the calculation of accurate three dimensional elbow angles. The positions of the two elbow epicondyles relative to the position of the upper arm triad are determined in “pointer calibration trials” which are collected prior to data collection (see Figure 4, which shows the position of the right arm lateral epicondyle being recorded). During data capture of the bowling action the position of each epicondyle relative to the upper arm triad is reconstructed. The line formed between joining the two epicondyles represents the “flexion/ extension axis” of the elbow with the mid-point of this line representing the elbow joint center. Again, in a similar approach to that used for determining the knee joint flexion/extension axis in the lower limb model, we are currently piloting a mean helical axis approach (flexion/extension and pronation/ supination) to more accurately identify the elbow axes of rotation. It is our opinion that this approach will improve the accuracy of elbow data, particularly for bowlers with joint abnormalities, when compared to results obtained using an anatomically based transepicondylar axis.

Figure 4: The pointer method used for determining the lateral elbow epicondyle position relative to the upper arm triad.

(Picture courtesy of Tom Rovis-Hermann, West Australian newspaper)

Wrist Center and Hand Motion

Calculation of the wrist joint center relies on the presence of two wrist markers placed on the styloid process of the radius and distal end of the ulna (see Figure 4, right forearm). During an initial static calibration trial the positions of the wrist markers are recorded relative to the position of another triad placed on the forearm, slightly superior to the wrist (Figures 3 & 4). Notice how the orientation of this triad is across the lower third of the forearm rather than vertically down as with the upper arm triad. This placement allows for more accurate pronation/supination data. The wrist markers may be removed prior to bowling as seen in Figure 3. During bowling trials hand motion is determined from two markers placed on the head of the 1st and 5th metacarpal.

Data collection procedures

Trials are typically collected until six successful deliveries have been recorded. To ensure that trials collected are as representative of the “match-environment” as possible, the following conditions needed to be met:

•  Ball needed to be of “good length” and spin in the appropriate direction (for a spin bowler)

•  Ball delivery speed needed to be over 90% of match speed.

•  An independent expert cricket coach agreed that the ball was delivered in the appropriate manner (similar to match deliveries).

•  At least five trials of each delivery variation should be collected for analysis.

Summary

This article is intended to provide a non-technical overview of the methods used by the UWA biomechanics group to assess fast and spin bowling techniques in cricket. In recognising the current limitations of the model we intend to continue to work on improving the accuracy of this model such that it becomes the standardized model adopted by ICC authorized testing centers throughout the world. We further contend that standardized biomechanical models and testing procedures are essential if the biomechanics community is going to provide consistent, accurate and meaningful results to national and international cricketing bodies concerning the legality of bowling actions.

The new Vicon system with 4 UWA's upper limb biomechanical model is currently being used for a number of other research projects:

. Tennis serve . Gymnastics tumbling . Field hockey flick

. Bowing in cello playing . Upper limb movements in cerebral palsy

Figure 5: The WA Symphony Orchestra's principal cellist, Rod McGrath undergoing testing. (West Australian newspaper)

References

Portus, M., Mason, B., Rath, D. & Rosemond, C. (2003). Fast bowling arm actions and the illegal delivery law in men's high performance cricket matches. Science and Medicine in Cricket. R. Stretch, T. Noakes & C. Vaughan (Eds), Com Press, Port Elizabeth, South Africa: 41-54.

Lloyd, D.G., Alderson, J. and Elliott, B.C. (2000). An upper limb kinematic model for the examination of cricket bowling: A case study of Muttiah Muralitharan. Journal of Sports Sciences, 18 (12):975-982.

Piazza, S., Okita, N., Cavanagh, P. (2001). Accuracy of the functional method of hip joint center location: effects of limited motion and varied implementation. Journal of Biomechanics, 34 (7): 967-973.

Richards, J. (1999). The measurement of human motion: A comparison of commercially available systems. Human Movement Science, 18: 589-602.

Besier T.F., Sturnieks, D.L., Alderson, J.A, and Lloyd D.G. (2003). Repeatability of gait data using functional hip joint center and knee helical axis. Journal of Biomechanics, 36 (8): 1159-1168.

HM&ES Laboratory Structure

The motion capture laboratory at HM&ES houses twelve cameras capable of operating at 1000 Hz. These cameras are strategically placed in the laboratory, so that they define a calibration space of 5m (long) x 2.5m (wide) x 3m (high). This permits the full delivery action and initial ball flight to be captured. The laboratory opens onto an oval, which allows a full run-up followed by a delivery onto a full-length cricket pitch.

As previously mentioned the methods used to define joint centers and axes of rotation are crucial for accurate kinematic data. The following joint center and joint axes definitions were employed in the HM&ES Upper Limb Model.

Shoulder Joint Center and Upper Arm Motion

The shoulder joint center (SJC) relies on accurate placement of an acromion marker positioned centrally on the acromion process (see Figure 2, left and right acromion markers), an anterior shoulder marker (see Figure 2 left anterior marker) and a posterior shoulder marker. The anterior and posterior markers are placed such that a line drawn between the two markers would bisect an approximated shoulder joint center (inferred frontal plane shoulder axis of rotation). The SJC is then calculated from where a vertical line dropped from the acromion intersects the line formed between the anterior and posterior shoulder markers. Research currently being conducted by the School of HM&ES is attempting to assess the accuracy of this method in locating the SJC using three dimensional computer tomography (CT) (Figure 2). Furthermore, in a similar method to that used for determining the hip joint center in the UWA lower limb model (Besier et. al., 2003), the biomechanics group is currently piloting the use of functional joint center approach for determining the location of the SJC (Piazza et al., 2001).