Archery robot and string fingers: Wakatabe 2015

The authors built a robot that shoots a recurve bow and fitted it with a new two-finger mechanism able to open its grip on the string at a controlled speed or to let the string slip out under its own tension alone. The aim was to reach a flex in the arrow equal to the one a person produces, since the previous robot produced twice as much and therefore did not reproduce real flight.

With a high-speed camera one metre in front of the bow, the amplitude of the arrow's lateral flex came out at 7.07 millimetres with the fingers opening at 1.125 degrees per second, 7.44 at 3.375 degrees per second, 7.30 at 5.625 degrees per second and 6.94 for the release by slipping, against 6.86 millimetres for an international-level woman archer and 14.8 for the earlier mechanism. The frequencies are close to one another: from 83.7 to 88.5 hertz across the robot's four conditions, 86.8 for the archer and 84.8 for the old mechanism. The standard deviation of the amplitude, that is the repeatability from shot to shot, runs from 0.18 to 0.23 millimetres for the robot and sits at 0.34 for the archer: the robot is steadier than a person.

One result went against expectation. The authors expected amplitude to grow with the speed at which the fingers open, and instead it came out larger at the intermediate speed than at the highest one: they therefore suppose that somewhere between those two values lies a speed at which amplitude peaks, and that beyond that threshold amplitude falls again, only to rise once more at very high speeds, where the velocity component perpendicular to the string plane becomes large.

The robot has six parts: the bow mount, the finger mechanism, the drawing unit, the elevation adjustment, the body and the control unit. The mount grips the bow's handle directly, with the grip removed, through a contact piece set on a rotation axis and a linear guide, so that the bow keeps two degrees of freedom, fore-and-aft translation and roll. This is the part redesigned for this piece of work: on the previous robot repeated shots produced play in the mount, and after a few shots the attitude was no longer held.

The new mechanism has two plastic fingers, with a worm and worm wheel on the fingers' rotation axis. That pairing stops the string tension from opening the fingers by itself and allows two modes: motorised opening at constant speed, tested at three values (motor at 30, 90 and 150 revolutions per minute, that is 1.125, 3.375 and 5.625 degrees per second), and static opening up to the minimum slipping angle, where the string comes off through friction. The drawing unit is a ball-screw mechanism with a servo motor and a linear guide, which carries the finger mechanism and draws the string with the arrow nocked; the elevation angle is set with a screw and read with an inclination sensor; the control unit is a motor controller with a laptop, set up for the length of the bow and the draw length.

The flex is measured by a high-speed camera placed one metre in front of the launching device, filming the arrow in flight from above. The bow and arrow used are listed in a table of specifications; the comparison is with shots from an international-level woman archer and with those of the earlier finger mechanism. The work was carried out with support from the Japanese national institute of sports sciences.

These are three pages of conference proceedings, without peer review, and the methods section is compressed. How many shots made up each condition is not stated, although the standard deviations reported imply a series of repetitions, and neither is the number of shots in the comparison with the archer. Of the archer nothing is known beyond her level, and one person does not stand in for anyone else's release technique: the flex of the arrow depends on how the string leaves the fingers, which is the most individual thing in the shot.

The measurement concerns the amplitude and frequency of the lateral flex one metre from the bow, filmed from above. There are no data on the point of impact, on group size, on arrow speed or on the behaviour further down the flight, so the resemblance between robot and person is claimed on one parameter rather than on the whole shot. The test was run with a single bow and a single arrow configuration, and the robot's greater repeatability should be read for what it is: a property of the machine, useful as a test bench, not a model of what an archer does. The hypothesis about a finger-opening speed at which amplitude peaks remains a hypothesis, since no other speeds were tested between 90 and 150 revolutions per minute and nothing above 150 was tried.

This study is for anyone who wants to measure arrows rather than for anyone who wants to shoot better. The problem it solves belongs to the test bench: a shooting machine fires far more consistently than a person, but if its release does not bend the arrow the way a hand bends it, what is measured in flight is not the flight of a shot arrow. Here the new mechanism brings the flex amplitude down from 14.8 to somewhere between 6.94 and 7.44 millimetres, against the 6.86 of an international-level archer, at similar frequencies: from this point on, a comparison between vanes, points or shafts run with the robot is more defensible than it was.

For archers themselves one piece of context stays interesting, and it is the spread. In the archer the flex of the arrow varied from shot to shot with a standard deviation of 0.34 millimetres, in the robot between 0.18 and 0.23: on a quantity of seven millimetres, then, the variability of a world-class archer is about 5%, and a machine halves it. It is a measure of how repeatable, and how far from repeatable, a finger release is. The study measures no scores and compares no techniques, so there is nothing here to take onto the shooting line.