Bayesian vs. least-squares inverse kinematics: Simulation experiments with models of 3D rigid body motion and 2D models including soft-tissue artefacts

A penalty method for constrained multibody kinematics optimisation using a Levenberg-Marquardt algorithm

An alternative method for solving constrained multibody kinematics optimisation using a penalty method on constraints and a Levenberg-Marquardt algorithm is proposed. It is compared to an optimisation resolution with hard kinematic constraints. These methods are applied to two pairs of experiments and models. The penalty method was at least 20 times faster than the optimisation resolution while keeping similar reconstruction errors and constraints violation. The potential of the method is shown to accurately solve the multibody kinematics optimisation problem in a reasonable amount of time. A computational gain lies in implementing this resolution with a compiled and optimised program code.

Comparing the performance of Bayesian and least-squares approaches for inverse kinematics problems

Bayesian inference has recently been identified as an approach for estimating a subjects' pose from noisy marker position data. Previous research suggests that Bayesian inference markedly reduces error for inverse kinematic problems relative to traditional least-squares approaches with estimators having reduced variance despite both least-squares and Bayesian estimators being unbiased. This result is surprising as Bayesian estimators are typically similar to least-squares approaches unless highly informative prior distributions are used. As a result the purpose of this work was to examine the sensitivity of Bayesian inverse kinematics solutions to the prior distribution. Our results highlight that Bayesian solutions to inverse kinematics are sensitive to the choice of prior and that the previously reported superior performance of Bayesian inference is likely due to an overly informative prior distribution which unrealistically uses knowledge of the true kinematic pose. When more realistic, 'weakly-informative' priors, which do not use the known kinematic pose are used then any improvements in estimator accuracy are minimal when compared to the traditional least squares approach. However, with appropriate priors, Bayesian inference can propagate uncertainties related to marker position to uncertainty in joint angles, a valuable contribution for kinematic analyses. When using Bayesian methods, we recommend researchers use weakly-informative priors and conduct a sensitivity analysis to highlight the effects of prior choice on analysis outcomes.

Total Least-Squares Determination of Body Segment Attitude

To examine segment and joint attitudes when using image-based motion capture, it is necessary to determine the rigid body transformation parameters from an inertial reference frame to a reference frame fixed in a body segment. Determine the rigid body transformation parameters must account for errors in the coordinates measured in both reference frames, a total least-squares problem. This study presents a new derivation that shows that a singular value decomposition-based method provides a total least-squares estimate of rigid body transformation parameters. The total least-squares method was compared with an algebraic method for determining rigid body attitude (TRIAD method). Two cases were examined: case 1 where the positions of a marker cluster contained noise after the transformation, and case 2 where the positions of a marker cluster contained noise both before and after the transformation. The white noise added to position data had a standard deviation from zero to 0.002 m, with 101 noise levels examined. For each noise level, 10 000 criterion attitude matrices were generated. Errors in estimating rigid body attitude were quantified by computing the angle, error angle, required to align the estimated rigid body attitude with the actual rigid body attitude. For both methods and cases, as the noise level increased the error angle increased, with errors larger for case 2 compared with case 1. The singular value decomposition (SVD)-based method was superior to the TRIAD algorithm for all noise levels and both cases, and provided a total least-squares estimate of body attitude.