The classic iterative Newton-Euler method for inverse dynamics applied to calculating net joint torques in biomechanics analysis has a number of drawbacks. Many sources of error including imprecision in video motion capture data measurements can lead to significant errors in calculated net joint torques. Adding ground reaction force data overconstrains the solution. This study examined the effectiveness of various inverse dynamics analysis methods on a full body analysis of the standing long jump motion. These methods included variations in which equations for segments from the linksegment model were removed to relieve over constraint. Also considered were analysis methods applying least squares optimization, which included all the measured data weighted in a least squares sense to fit to an overconstrained system.

Motion capture data of 48 total standing long jump trials were collected and analyzed. Conventional iterative solutions with and without including measured ground reaction forces, and least squares optimized inverse dynamics solutions were derived and applied to the kinematic data in a 2-dimensional, seven-segment, linked segment model of the full body. Net joint torques were calculated at six joints for a 1.5 s period immediately prior to take-off of each standing long jump, and joint power and total work performed at each joint was calculated over the entirety of each jump. The optimized least squares solution was shown to be very similar to the conventional iterative solution using ground reaction forces and removing the equations of motion at the trunk segment. Net mean torques at the elbow and shoulder were highly variable.