Visual3D has two distinctive approaches to computing the position and orientation of a segment. The first approach is the Six Degrees of Freedom method, which determines the position and orientation of segments independently from one another based solely upon measurements of that segment. The second approach is the Inverse Kinematics method, where segments form a hierarchical linked chain with joint properties that define the “connection” between segments.

The Six Degrees of Freedom method (6 DOF) is also referred to as Segment Optimization, since segment positions and orientations are solved for on a per-segment basis.

For marker-based motion capture, a set of 3 or more markers attached to a rigid segment is used to track the movement of the segment and at each frame of data the pose (position and orientation) of the segment is estimated.

The method's name refers to the fact that each segment (or each joint) is considered to have 6 variables that describe its pose (3 variable describe the position of the origin, 3 variables describe the rotation about each of the principal axes of the segment)

An alternative to the 6 DOF solution is to define joints (e.g. explicitly state which segments are connected by a joint) and to specify the properties of all joints. Because the targets used to track the segments are often subject to measurement error and soft tissue artifact, motion about some of the degrees of freedom maybe much larger than the motion that would be realistically possible.

The Inverse Kinematics method (IK) is also referred to as Global Optimization, since segment positions and orientations are solved in order to minimize a global loss function while respecting these joint constraints.

Lu and O’Connor (1999) described a global optimization process where physically realistic joint constraints can be added to the model to minimize the effect of the soft tissue and measurement error. Lu and O’Connor termed this process Global Optimization while other inside the biomechanics community prefer the term “Inverse Kinematics.” Visual3D uses the term Inverse Kinematics but the underlying calculations are based upon the Lu and O’Connor technique.)

Many practitioners are interested in the questions: How close are the 6 DOF and IK methods in terms of data and which one is correct?

These questions are almost impossible to answer because the correct answers will vary tremendously due to difference in the quality of the lab data, the type of movement, the amount and type of soft-tissue movement and the validity of the joint constraints used for the IK method.

In Lu and O'Connors landmark paper, they found that IK (which they refer to as Global Optimization) produced better results then 6 DOF (which they refer to as Segmental Optimization). However Lu and O'Connor's test was based on a simple model of soft-tissue error and assumed the joint constraints were an accurate representation of the subject (all joints were ball joints).

In general I find that a visual inspection of the data in Visual3D will give you a good clue of whether IK is useful or not. If looking at the data in Visual3D you see a lot of joints disarticulating then IK will generally be a good idea. (For example I have looked at the upper extremities in baseball pitching and golf and you often see the elbow blow apart and IK helps this sort of data considerably.)

Lecture notes on pose estimation], including 6 DOF and Inverse Kinematics (e.g. Segment Optimization and Global Optimization).

Allan Carman] has prepared an interesting expose on marker sets and pose estimation algorithms.

While these opinions are Allan's, not ours, we highly recommend reading the information.

If you have any questions related to Allan's document and how things are implemented in Visual3D, you will probably find it in our wiki or text book chapters, but feel free to contact us about specific issues.

Leardini A, Belvedere C, Nardini F, Sancisi N, Conconi M, Parenti-Castelli V (2017) Kinematic models of lower limb joints for musculo-skeletal modelling and optimization in gait analysis. J Biomech. 2017 Sep 6;62:77-86. doi: 10.1016/j.jbiomech.2017.04.029. Epub 2017 May 22.

Abstract

Kinematic models of lower limb joints have several potential applications in musculoskeletal modelling of the locomotion apparatus, including the reproduction of the natural joint motion. These models have recently revealed their value also for in vivo motion analysis experiments, where the soft-tissue artefact is a critical known problem. This arises at the interface between the skin markers and the underlying bone, and can be reduced by defining multibody kinematic models of the lower limb and by running optimization processes aimed at obtaining estimates of position and orientation of relevant bones. With respect to standard methods based on the separate optimization of each single body segment, this technique makes it also possible to respect joint kinematic constraints. Whereas the hip joint is traditionally assumed as a 3 degrees of freedom ball and socket articulation, many previous studies have proposed a number of different kinematic models for the knee and ankle joints. Some of these are rigid, while others have compliant elements. Some models have clear anatomical correspondences and include real joint constraints; other models are more kinematically oriented, these being mainly aimed at reproducing joint kinematics. This paper provides a critical review of the kinematic models reported in literature for the major lower limb joints and used for the reduction of soft-tissue artefact. Advantages and disadvantages of these models are discussed, considering their anatomical significance, accuracy of predictions, computational costs, feasibility of personalization, and other features. Their use in the optimization process is also addressed, both in normal and pathological subjects.

Schmitz A1, Buczek FL, Bruening D, Rainbow MJ, Cooney K, Thelen D. (2015) Comparison of hierarchical and six degrees-of-freedom marker sets in analyzing gait kinematics. Comput Methods Biomech Biomed Engin. 2015 Mar 24:1-9. [Epub ahead of print]

The objective of this study was to determine how marker spacing, noise, and joint translations affect joint angle calculations using both a hierarchical and a six degrees-of-freedom (6DoF) marker set. A simple two-segment model demonstrates that a hierarchical marker set produces biased joint rotation estimates when sagittal joint translations occur whereas a 6DoF marker set mitigates these bias errors with precision improving with increased marker spacing. These effects were evident in gait simulations where the 6DoF marker set was shown to be more accurate at tracking axial rotation angles at the hip, knee, and ankle.

PudMed]