Ultrasound sensors for archery: Loke 2009 study

Rather than a study, the paper describes an instrument: a low-cost system, built at Monash University in Malaysia for the National Sports Institute, that uses ultrasound sensors to measure how much the bow moves while the archer aims. Four sensors, strapped to the bow and the stabiliser, measure the distance to two reference surfaces, one horizontal and one vertical. From these distances the software works out the bow's displacement in millimetres and its rotation about two axes, and for each shot it computes a variance that the authors use as an index of stability.

The paper contains no data from archers: no sample, no measured shots, no comparison with another instrument. The figures it gives concern the system itself: the smallest change in distance it can pick up is about 0.4 mm, and it records roughly 18 samples per second, a rate the authors themselves consider low (their target is more than 50).

The sensors are four Parallax Ping units, with a range of 2 cm to 3 metres, connected to a Basic Stamp II microcontroller that talks to the computer over the serial port; the software is written in LabVIEW. Each sensor measures the round-trip time of the pulse, and the distance is that time multiplied by the speed of sound and halved. Two sensors work on one axis and two on the other: the spacing between the two sensors of a pair must stay fixed, because the angle calculation depends on it, whereas their exact position on the bow does not matter. The signal is cleaned with a moving average over blocks of five samples, which replaces outlying values.

The interface has five tabs: session details, the real-time display, post-session processing with the variance of each attempt, motion playback and a comparison between two sessions. Recording starts automatically when the bow rises above a set height and the signal settles, which the software takes to mean that the archer is aiming.

The system was not compared with a reference instrument such as an optical motion system, so there is no way of knowing how accurate its displacements and angles are on the bow of an archer actually shooting. The authors acknowledge two weaknesses: at 18 samples per second fast movements are missed, and the start and end of aiming are poorly identified, so each attempt also contains samples from the set-up and from after the shot. Ultrasound is affected by temperature, humidity and air pressure, by electronic interference and by crosstalk between sensors; the paper does not say where on the range the surfaces that reflect the pulses are placed, nor whether the sensors alter the balance of the bow.

There is also a gap between the name and the measurement: the system is meant to assess the stability of the hand, but it records the movement of the bow, and the variance chosen as an index is never related to where the arrow lands or what it scores.

For anyone who shoots, the paper changes nothing: it says neither how much archers' bows move while aiming nor whether a steadier bow scores more. Its value is mainly historical, since it shows that in 2009 a national institute was looking for a cheap way to measure aiming in training without loading the archer with body-worn sensors. At 18 samples per second there are about 55 milliseconds between one reading and the next: the system can follow slow drifts in the aim, not what happens at the release.

The most useful detail is the fix the authors propose for the measurement window: a wire connected across the two sides of the clicker, giving an electrical signal at the moment the arrow passes it. The same principle, an electrical signal tied to the clicker dropping, turns up in other studies that use it to synchronise force plates and cameras.