The efficacy of a golf robot for simulating individual golfer's swings

Golfers are used extensively for equipment testing and development, however, they are inconsistent in performance and they tire. Golf swing simulation devices overcome these shortcomings producing consistent repeatable swings and are now intrinsic instruments for golf research conducted by manufacturers governing bodies and academic institutes. The capability of artificial devices to perform human dynamic motions is undoubtedly improving; however, comparatively little scientific effort has been invested to ensure that the swing motions provided by these mechanical devices are representative of golfers' swings. An extensive 3D kinematic study of golfers' swings was undertaken to determine joint positions representative of a double pendulum model of the swing along with ball launch and shaft deflection data. A technique for generating golf robot motion commands from the kinematic swing data was then developed using interpolation algorithms to increase the continuity of command derivatives. Six golfers' swings with distinctive swing performance characteristics were replicated using a commercially availablem odem golf robot and a bespokem otion control system. Quantitative data captured from 20 golfers were compared using gender and skill classifications. All golfers were found to perform consistent swing motions and the kinematic sequencing of joint segments supported the summation of speed principle. Higher skill rated golfers were found to perform swings with greater consistency, faster clubhead speeds at impact, faster down swings and greater delayed wrist releases. The research has shown that the new robot control system in conjunction with 3D kinematic swing data and a feed-forward control enables in dividual golfer's swings to be replicated with both high levels of accuracy and repeatability. The capability for golfers' swings to be replicated accurately using a mechanical device provides new impetus for both clubfitting methodologies and optimisation of swing technique. However, the boundary constraint of the club at the robot's gripping mechanism was found to provide poor representation of golfers' grip, and consequently, the first bending mode of the club was excited during simulations and resultant shaft deflection and ball launch data correlated less well with the golfers' data. Development of the club gripping mechanism has been identified as an area for future research as the benefits of the new robot control system will only be fully realised if this interface is improved.