# Changelog

% A rendered version of the CHANGELOG is avaible here:

%    https://anyscript.org/ammr/beta/changelog.html

(ammr-3.1.0-changelog)=

## AMMR 3.1.0 (2025-03-31)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.15094590.svg)](https://doi.org/10.5281/zenodo.15094590)

[![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-8.1.0-yellowgreen)](https://www.anybodytech.com/resources/customer-downloads/)

### 🔧 Changed:

*   Modified the Human-Ground residual implementation in MoCap models to use

    rotational measures configured for angular velocities. This change enhances

    the robustness of the residuals and makes the residual output easier to

    interpret geometrically, without affecting the results.

*   Updated the insertion and origin points of the Gluteus Medius to ensure

    correct moment arms for external rotation in specific postures. The posterior

    Gluteus Medius now twists inside the anterior part and attaches more

    anteriorly on the femoral trochanter.

*   The wrapping surface for muscles at the wrist has been changed from a cylinder

    to an ellipsoid. This lowers the risk of muscle via points penetrating the

    wrapping surface in postures involving both flexion/extension and

    abduction/adduction at the wrist.

*   Excluded force plates from the parameter identification study, as they were

    unnecessary. This change may slightly speed up the parameter identification

    process.

*   The coordinate system used when scaling the shank segment has been aligned

    with same reference frame used when scaling the thigh segment, instead of

    using the anatomical frame of the shank. Now both segments use the direction

    given by the femoral epicondyles as the lateral direction. This change

    ensures consistent scaling across the knee joint. 

### ➕ Added:

*   Introduced a new class template,

    [`CreateCoMRefNode`](#Utilities.center-of-mass.createcomrefnode.createcomrefnode),

    for generating a reference node at the center of mass of a segment, aligned

    with its principal axes of inertia.

*   Added the `TRI(N)` macro to create a lower triangular (NxN) matrix.

*   Added the `TOTAL_POLYLINE_LENGTH(P)` macro to compute the total length of a

    polyline defined by a set of points.

### 🩹 Fixed:

*   Corrected the PelvisGround rotation drivers in the {ref}`Free Posture Static

    example<example_freeposture>`. The X and Z rotation drivers were previously

    mixed up and are now properly assigned.

*   Resolved an issue that prevented disabling the drawing of marker arrows in

    `CreateMarkerDriverClass` within MoCap models. The search string in

    `Main.ModelSetup.Views.All_MarkerArrows.Objects` has been updated to correctly

    identify the arrow drawing objects.

*   Fixed a spelling typo in the disc stiffness BM control statements. The

    `BM_TRUNK_*_DISC_STIFNESS` parameters have been renamed to `BM_TRUNK_*_DISC_STIFFNESS`.

    The misspelled parameter names have been deprecated.

*   The data points for the femoral epicondyles with respect to the shank in the

    neutral posture have been corrected. These values were not updated when the

    knee axis in shank was redefined in TLEM 2.2 and AMMR 3. This change has no

    influence on the default AnyBody Models, but it fixes an issue if the model

    was configured to use the {bm_constant}`special 'bony landmark' defined <_JOINT_TYPE_BONY_LANDMARK_>`

    joint axis configuration.

### Removed:

*   Removed the python hooks to check for modified AMMR folder. This feature was

    not used and has been removed to simplify the AMMR codebase.

(ammr-3.0.4-changelog)=

## AMMR 3.0.4 (2024-07-02)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.12592455.svg)](https://doi.org/10.5281/zenodo.12592455)

[![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-8.0.4-yellowgreen)](https://www.anybodytech.com/resources/customer-downloads/)

### 🩹 Fixed:

* Fixed an issue in the {ref}`Bike Model example <example_bikemodel>` that

  included thoracic support in the full body model also.

* Fixed an issue with the visualization of the marker coordinate system arrows. They now

  correctly update while running running parameter identification in the MoCap models.

(ammr-3.0.3-changelog)=

## AMMR 3.0.3 (2024-06-10)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.11191711.svg)](https://doi.org/10.5281/zenodo.11191711)

[![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-8.0.3-yellowgreen)](https://www.anybodytech.com/resources/customer-downloads/)

### 🩹 Fixed:

* Fixed a penetration warning for the pectoralis muscles when the thoracic

  segments are scaled very non-uniformly. The fix involves a small (5 deg)

  adjustment to the orientation of the pectoralis wrapping surface.

* Fixed default limits in [Range of Motion limits](#Utilities.Kinematic-limits.RangeOfMotionLimits_template.any)

  class template. The default limits for PelvisThoraxExtension, PelvisThoraxLateralBending,

  Right/LeftWristFlexion had the upper and lower limits flipped. This is now fixed and a 

  check is added in the class template to catch this kind of error.

* Fixed an issue where blinking Windows would appear when loading models stored

  in git repositories. This issue was caused by a python subprocess that queried

  the git repository for branch information.

* Updated the foot marker position in the Xsens protocol to accommodate the

  changes made to the foot anatomical frame in the TLEM 2.2 leg model. The

  R/LTOE and R/LTOE2 markers have been moved upwards by 1.5 cm.

(ammr-3.0.1-changelog)=

## AMMR 3.0.1 (2024-02-13)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.10803883.svg)](https://doi.org/10.5281/zenodo.10803883)

[![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-8.0.1-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-8-0-1-11638_x64/)

This minor release of AMMR contains no updates to the models compared to the major AMMR 3.0.0 released the previous month. It only has a few minor changes to the documentation and new tolerances in the FDK models. 

### 🔧 Changed:

* Finer FDK tolerances for the {ref}`Knee Simulator example

  <example_kneesimulator>`, to take avantage of the improved robustness of the

  AnyBody FDK solver in version 8.0.1. 

(ammr-3.0-changelog)=

## AMMR 3.0 (2024-02-13)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.10527958.svg)](https://doi.org/10.5281/zenodo.10527958)

[![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-8.0.0-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-8-0-0-11540_x64/)

### ✨ Highlights 

* Integrated pelvis, lumbar, and thorax models into a consistent trunk model in

  preparation for a full thoracic model in [AMMR 4](https://anyscript.org/ammr/beta/body/thoracic_model.html).

* Updated the leg model (TLEM 2.2) with improvements to the ankle complex, 

  shank orientation, and alignment of the patella tendon.

* Added new wrapping surfaces to the Achilles tendon around the 

  ankle in the TLEM 2.2 leg model.

* Made numerous improvements and fixes to the Body models, MoCap models,

  and application examples.

:::{Warning} Care must be taken when porting existing MoCap models to the 

new AMMR. Changes to the pelvic tilt, and changes to the foot anatomical

frame can cause MoCap markers on Pelvis and feet to have moved slightly.

If migrating models from AMMR 2.4, be sure to check markers on

pelvis and the foot segments.  The Y component may need to be adjusted to 

achieve same marker position and joint angle output.

:::

### 🩹 Fixed:

* Fixed a couple of bugs in the {ref}`Seated Human model <example_seatedhuman>`:  

  - Reactions for subtalar joint driver are switched off.

  - GHDriverLeft was using velocity value from GHExternalRotation instead of GHAbduction.

    This is now fixed.

* The `Main.ModelSetup.CreateVideo` operation was missing in some of the MoCap

  examples. This has been fixed. If you have this problem update the

  `CreateVideo.any` file in your application. 

* Fixed a wrapping problem with the posterior deltoid muscle in the 2-parameter

  shoulder calibration. 

* Fixed a bug in the LegPressMachine example that caused the model view to zoom

  to infinity.

* Improved the robustness of muscle recruitment for abdominal muscles (`buckle

  support`) by increasing the strength of recruited reactions on buckle segment

  balance.

* Updated scaling functions (`Scale`) for segments in trunk model (lumbar,

  thoracic and cervical) to account for pelvis segment morphing into leg pelvis

  coordinate system using `BM_LEG_TRUNK_INTERFACE` setting.

* Fixed a bug in the `AnatomicalFrame` of TLEM 2.2 foot and talus that caused the

  ARel to be constructed from unscaled landmarks.

* Fixed small inaccuracy in the mass scaling of the shank. It did not account for the

  mass of the patella segment which uses the same scaling coefficient as the shank.

* Visualizaiton of muscles in AMMR now use the builtin `.viewMuscle` drawing objects 

  in the muscle class. Hence, all manually created `AnyDrawMuscle` objects, usually named

  `DrwMus` have been removed. 

* The kinematic start guess for left patalla joint has been adjusted to improve

  robustness of the kinematic solver. This does not changed the result of

  models. 

### ➕ Added:

* Added two small helper code macros:

  `NON_UNIQUE_VALUES()`/`NON_UNIQUE_POINTERS()` for finding duplicate

  (non-unique) entries in arrays of values and pointers.

* Added new set of hip joint measures (`Interface.Right.HipISB`) that strictly

  follow the International Society of Biomechanics' (ISB) definitions. These

  measures are similar to existing angles but do not have zero hip-flexion in

  neutral position due to different definition of pelvis coordinate system with

  respect to a plane defined by ASIS-PSIS points.

* Segments in trunk model (lumbar, thoracic and cervical) now explicitly define

  a `ScalingNode` node indicating coordinate system for segment scaling.

* Improved error message when MoCap markers in marker protocol are missing from

  C3D file.

* Add an option to override the default kinematic joint limits in the MoCap

  framework

### 🔧 Changed:

* The trunk joint reaction force in `SelectedOutputs`

  is now expressed in new reference nodes that are aligned with the vertebral body.

  Therefore, the names of the outputs now include CompressionForce and 

  ShearForce. The new reference nodes are created inside the joint node on the proximal

  segment. The implementation from AMMR 2.4 can be enabled by using the compatibility

  switch: `#define BM_COMPATIBILITY_24_TRUNK_SELECTED_OUTPUTS_JRF ON`

* Adjustments to pelvic tilt have changed the position of MoCap markers on the

  pelvis segment. This mostly affects markers with hardcoded positions.

  Particularly PSIS markers furthest posterior from the origin between ASIS.

  When migrating from AMMR 2.4 models without adjusting marker positions, PSIS

  markers may move 2 cm lower.

  - Most [non-MoCap model examples](#example-gallery)

    have been updated to account for the new pelvic tilt value. Either by using the 

    `AnatomicalFrameTrunk` reference frame in the pelvic driver or by adjusting the 

    Pelvis-Thorax Extension slightly.

* The Twente Lower Extremity Model (TLEM) 2 leg model has several updates and is

  now designated as [TLEM 2.2](#TLEM2-v2.2).

  - The ankle complex and the knee joint axis have been redefined to compensate for

    the non-neutral position in which the cadaver was scanned. This improves the 

    alignment of the tibia, talus, and calcaneus bones. The updated knee joint ensures 

    that the patella tendon is straight in the neutral position. The net effect is 

    rotation of the tibia about its long axis with the feet still pointing in the 

    same direction in the neutral position. 

  - Wrapping surfaces have been added to the Achilles tendon around the ankle in

    the TLEM 2.2 leg model, giving a more even ratio of moment arms between

    the soleus and gastrocnemius muscles. Hence, gastrocnemius is recruited

    less, especially during downhill walking and stair descent, solving the

    tendency of the model to over-predict the knee contact forces at toe off.

    This is the first of a number of improvements to the leg model by Dr. Enrico

    De Pieri, who is working on a publication on improvements and validation of

    the TLEM 2 leg model. 

  - The foot and talus models in TLEM 2.2 leg model have several updates in

    preparation for the release of advanced multi-segment foot models in the

    future:

    - The coordinate system of the talus is updated to be coincident with the foot

      coordinate system. A new reference node, `TalusCompatibilityFrameAMMR24`,

      is created in the talus segment to be consistent with the previous

      coordinate system for backwards compatibility.

    - Subtalar and ankle joint parameters for the talus have been updated to be

      expressed in the new coordinate system. The implementation of the joints

      is still consistent with the previous implementation.    

    - The anatomical frames of the foot and talus are updated and defined using

      bony landmarks on the foot. This updates the neutral position of the foot

      and talus. This will also update the ankle plantarflexion and subtalar

      joint angles. 

    - The malleoli coordinates in the foot coordinate system have been fixed to

      match the malleoli on the shank in the neutral position.

    - The model tree has been updated. The talus segment is moved inside the

      foot segment. For backwards compatibility, a pointer to the talus segment

      still exists outside the foot segment.

  - The implementation of the TLEM model has been split into two separate folders,

    `LegTLEM/` and `LegTLEM1/`. All references to TLEM 1 code have been removed in

    the code implementing the TLEM 2 model. 

  - The TLEM 1 model now triggers a deprecation warning suggesting using TLEM 2.

* The InverseDynamics analysis in MoCap models now uses the first step of Marker

  tracking as a start guess (initial conditions)  

  for the kinematic solver. This improves robustness in certain cases.

* The marker protocol in the {ref}`"ADL gait (MoCap model)"

  <sphx_glr_auto_examples_MoCap_plot_ADL_Gait.py>` was updated/tweaked to

  improve the robustness of the model. Some special exceptions were added to

  trials where we discovered that some markers were swapped by accident in the

  original open access dataset.

* Scaling laws defined by the `BM_SCALING` setting have been updated to calculate

  offsets between different scaling regions and apply them at load time. Scaling

  remains the same, but users can now create scaling functions that account for

  offsets between regions with different scaling.

* Glenoid reaction forces are now expressed in the coordinate system of the glenoid

  cup instead of the general scapula coordinate system. Variables

  `GlenoHumeral_DistractionForce`, `GlenoHumeral_InferoSuperiorForce`,

  `GlenoHumeral_AnteroPosteriorForce` represent three directions given by the

  glenoid cup. Variables may differ slightly even if the force remains the same.

* Resolved inconsistencies in arm muscle parameters. The same underlying parameters

  are now used for both simple and 3-element muscle models.

* MoCap marker protocols: Users must now explicitly specify a coordinate system

  relative to which markers are placed on the segment using the 

  [`PlaceMarkerAt`](#MoCap.Markers.CreateMarkerDriverClass.CreateMarkerDriver:PlaceMarkerAt) 

  argument in the class template. Previously, this defaulted to the `AnatomicalFrame` of

  the segment.

* The organization of segmental scaling functions was reworked and now configured

  slightly simpler. This modification is expected not to affect users working

  with default and non-default scaling laws and patient-specific morphing.

  Additionally, all spinal segments share a scaling reference frame.

* We updated how the class `AnySurfCylinderFit` is used in the {ref}`Knee

  Simulator example <example_kneesimulator>`. The deprecated way of using the

  `AnySurfCylinderFit` has been changed. Since `AnySurfCylinderFit` is now a

  subclass of `AnyRefNode`, it should no longer be nested inside another

  parent reference frame to provide the pos `sRel` and orientation `ARel`. This

  fixes a number of deprecated warnings that would have triggered with AnyBody version 8. 

* Added the new `AnatomicalFrameTrunk` reference frame to the pelvis segment consistent

  with anatomical frames in the rest of the trunk model. All joint angles relative

  to the pelvis segment now use this frame. This aligns the neutral position of

  the model with the neutral position of the trunk dataset and results in more

  pelvic tilt in the neutral position, better reflecting reported values in

  literature. Existing `PelvisSeg.AnatomicalFrame` defined by the ASIS and pubic bony

  landmarks remains unchanged.

* Updated trend validation in ["Wilke Spine Pressure validation

  models"](example_wilkespinepressure) to reflect changes to the thoracic model

  and the pelvis frame. Although absolute pressures in the spine have changed,

  trends (relative changes in spine pressure between models) remain the same.

* Reduced discretization (number of elements) of the pectoralis major and

  semispinalis muscles for consistency with textbook anatomy. Pectoralis no

  longer has its origin on rib 2, and semispinalis no longer inserts on the

  lumbar region.

* Refactored the implementation of muscle parameters in the arm model. All

  parameters are now given as muscle volume, optimal fiber length, and tendon

  slack length. Physiological cross-sectional area (PCSA), previously a

  hardcoded parameter, is now an intermediate value used for calculating muscle

  strength from muscle volume and optimal fiber length.

  This new structure allows for overwriting the complete set of muscle volumes with

  an alternative dataset.

* The [`MarkerName`](#MoCap.Markers.CreateMarkerDriverClass.CreateMarkerDriver:MarkerName) 

  argument in the [`CreateMarkerDriver`](#MoCap.Markers.CreateMarkerDriverClass.CreateMarkerDriver)

  template for MoCap models is no longer necessary. The argument can still be used 

  if the marker class and data entry in the C3D file differ.

* Updated the implementation of muscle parameters section in models to use the new

  `??=` (optional assignment) operator introduced in AnyBody 7.4.1. This allows

  for direct overriding/redefinition of muscle parameters and volumes.

  For example, muscle volumes can be overridden with new data.

  ```{code-block} AnyScriptDoc

  Main.HumanModel.BodyModel.Right.Leg.ModelParameters.Muscles = {

     SoleusMedialis.MuscleVolume = 540; //ml

     SoleusLateralis.MuscleVolume = 450; //ml

  };

  ```

* Updated methods for distributing muscle volume among discretized muscle

  elements. This allows for different optimal fiber lengths among elements of a

  muscle. If one element has a smaller optimal fiber length (e.g., through

  calibration), the volume of muscle elements is redistributed, and PCSA remains

  constant across elements. 

  If the optimal fiber length of different elements is

  the same, this method yields the same result as before.

* Renamed the folders holding muscle models from `MusPar` to `MuscleModels` for

  better clarity.

* The `CameraClassTemplate.any` include file in model utilities has been renamed to

  [`VideoLookAtCamera.any`](#Video-tools.Video.VideoLookAtCamera.any) to have the same name as the class template it

  contains. 

### 🧹 Removed:

- The template `DEFAULT_PARAMETER_FOLDER` has been removed. It is no longer needed 

  since the same functionality is now provided by the `??=` operator.

- The three scaling laws `_SCALING_UNIFORM_EXTMEASUREMENTS_`,

  `_SCALING_LENGTHMASS_EXTMEASUREMENTS_`,

  `_SCALING_LENGTHMASSFAT_EXTMEASUREMENTS_` have been deprecated. They are

  superseeded by the "Statistical Scaling Plugin" which cover the same

  functionality. The AMMR guide on scaling have been updated to reflect this

  change.

(ammr-2.4-changelog)=

## AMMR 2.4.4 (2023-04-03)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.7764841.svg)](https://doi.org/10.5281/zenodo.7764841)

[![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.4.4-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-4-4-8907_x64/)

This version of AMMR adds a number fixes and tweaks to the body models which

improves robustness of various kinematic and recruitment solvers. 

**Added:**

* Added kinematic joint angle limits in MoCap models for elbow and wrist joints

  to prevent the kinematic solver from finding postures that are physiologically

  impossible, such as bending the elbow backwards. These limits are active where

  marker tracking solver would occasionally find a local minima with

  unphysiological posture.

* Tables with body model configuration parameter in the AMMR documentation now

  contain links showing their options. For example, see the page on {doc}`Leg

  model parameter </bm_config/leg>`.

**Fixed:**

* Fixed an issue with wrist joint segment's load time position (start guess).

  This greatly improves kinematic robustness of all models with arms as it

  creates a 'universal-joint' mechanism for wrist movement.

* Further fixes made to pectoralis wrapping segment's robustness by optimizing

  initial load time positions to ensure model kinematics can more easily solve.

* Fixed missing `LoadParameters` operation in `LoadAndReplay` operation in MoCap

  examples.

* Fixed a problem with oblique muscles introduced in AMMR 2.4. A weak residual

  force was added to Y rotation of buckle to eliminate this problem.

* Corrected wrong order of nonlinear intervertebral disc stiffness polynomial

  coefficients (affects only those who used polynomial disc stiffness).

* Corrected small asymmetry in function for nonlinear intervertebral disc

  stiffness in coronal plane (affects only those who used polynomial disc

  stiffness).

## AMMR 2.4.3 (2023-01-27)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.7572879.svg)](https://doi.org/10.5281/zenodo.7572879)  [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.4.3-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-4-3-8899_x64/)

**Added:**

* Added a few utility helper class templates

  - `Template_OperationSaveValues`

  - `Template_OperationLoadValues` 

  - `Template_OperationUpdateValues`

  to make it easier to do common class operations without manually having to

  create the operations with macros.

  To create an operation which loads a file, do: 

  ```{code-block} AnyScriptDoc

  Template_OperationLoadValues LoadAnySetFile = {

     FileName= "MyFile.anyset";

  };

  ```

* A more helpful error message is now printed when MoCap markers in the marker

  protocol are missing the C3D file.

**Changed:**

* The load-time position of the box in the {ref}`BVH_BoxLift model

  <sphx_glr_auto_examples_Mocap_plot_BVH_BoxLift.py>` is now calculated using

  the position of the hands. Also,

  `Main.ModelSetup.EnvironmentParameters.GravityDirection` defined in the

  `box.any` file is now calculated automatically from

  `Main.ModelSetup.LabSpecificData.Gravity` defined in the `LabSpecificData.any`

  file. These changes should make the model more robust when dealing with

  different BVH files.

* It is no longer necessasry to supply the `MarkerName` argument in the

  CreateMarkerDriver template MoCap models. The argument can still be used if

  the marker name and the data entry in the c3d file are different.

* The references to muscle models in the joint muscles of the detailed hand have

  been renamed to avoid future naming conflicts.

* The "via-points" for the Psoas Major muscle have been adjusted to ensure that

  the muscles can better act the role of stabilizing muscle for the lumbar

  spine.

**Fixed:**

* The `Main.ModelSetup.CreateVideo` operation was missing in some of the MoCap

  examples. This has been fixed. If you have this problem please update the

  `CreateVideo.any` file in your application.

* Fixed a wrapping problem with the posterior deltoid muscle in the

  two-parameter shoulder calibration. 

* Fixed a bug in the LegPressMachine example that caused the model view to zoom

  to infinity. 

* The robustness of muscle recruitment with the abdominal muscles (`buckle

  support`) was futher tweaked by increasing the the strength of the recruited

  reactions on the balance of the buckle segment.  

## AMMR 2.4.2 (2022-07-08)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.6809697.svg)](https://doi.org/10.5281/zenodo.6809697) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.4.2-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-4-2-8845_x64/)

AMMR version 2.4.2 contains only a few minor changes and fixes compared to

version 2.4.1.

**Changed:**

* Changed defintion (sign) of the `SkullThoraxRotation` and

  `SkullThoraxLateralbending` variable in the `interface` folder. They are now

  consistent with similar difinitions in the rest of the spine.

* The location of the foot drawing generated by the `OptimizeBVH_Origin` class

  template for visualization of the target location and orientation of "LFOOT"

  and "RFOOT" has been updated to better match the expected actual foot

  location.

**Fixed:**

* Fixed the {ref}`Evaluate moments arms example

  <sphx_glr_auto_examples_Validation_plot_EvaluateMomentArms.py>` where a few

  branches of the latisimus dorsi, pectoralis major and trapezius muscles were

  not included. 

* Fixed a problem that caused the `OptimizeBVH_Origin` class template to fail to

  load if the same human segment was used more than one time in the same model.

## AMMR 2.4.1 (2022-05-30)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.6533249.svg)](https://doi.org/10.5281/zenodo.6533249) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.4.1-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-4-1-8830_x64/)

**New example models:**

- Added a model using a new class template to optimize the origin of BVH model.

  This model shows how to optimize the origin of the BVH model such that a

  target segment of the human model (Left/Right Foot/Hand) hits a known position

  and orientation in a given time interval while following the recorded motion

  from the trial. <{ref}`See more

  <sphx_glr_auto_examples_Mocap_plot_BVH_OptimizeOrigin.py>`>

**Fixed:**

* Fixed a small syntax error which caused the BVH example `S1/S01_Trial02/` to

  fail loading. 

* Fixed a problem with the "interface" morphing between the TLEM2 pelvis and

  Trunk Pelvis at the tip of the scarum bone. This bug was visible when using

  `#define BM_LEG_TRUNK_INTERFACE _MORPH_LEG_TO_TRUNK_`, (i.e. when using the

  trunk pelvis morphology with the TLEM2 model), which is not the default. 

* Fixed missing calibration for the new latissimus dorsi elements introduced in

  AMMR 2.4. This fix also ensures that all latissimus dorsi elements are

  calibrated in the same arm posture.

**Changed:**

* The anyscript implemenatation of the acromio-clavicuala, sterno-clavicular,

  gleno-humeral, and elbow joints have been refactored. They now use explicit

  AnyScript joint classes (e.g. `AnySphericalJoint`) instead of implementing the

  joints directly with measures and constraints. This simplifies the

  implementation, but also changes the structure of the model tree. However, it

  has no influence the kinematics of the model, which remains identical. 

  The new names of the explicit joint objects in `ShoulderArm.Jnt` are:

  * `SternoClavicularJoint` (-3 DOF)

  * `AcromioClavicularJoint` (-3 DOF)

  * `GlenoHumeralJoint` (-3 DOF)

  * `HumeroUlnarJoint` (-5 DOF)

  * `HumeroRadialJoint` (-1 DOF)

  * `ProximalRadioUlnarJoint` (-2 DOF)

  * `DistalRadioUlnarJoint` (-2 DOF)

The `HumeroUlnarJoint` is the elbow flexion extension, and together

`HumeroRadialJoint`, `ProximalRadioUlnarJoint` and `DistalRadioUlnarJoint` adds

5 constraints leaving the forearm rotation degree of freedom. 

**Added:**

* A few previously renamed nodes in pelvis were added back to improve backwards

  compatibility when loading old seating models. 

## AMMR 2.4 (2022-04-28)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.6471999.svg)](https://doi.org/10.5281/zenodo.6471999) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.4.0-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-4-0-8782_x64/)

**New example models:**

- Added a new trunk exoskeleton concept model for simulating assistance at the

  trunk in a box lifting task. This simple example shows how to apply assistive

  torque directly to the human model. <{ref}`See more

  <sphx_glr_auto_examples_Other_plot_ExoConceptTrunk_BoxLift.py>`>

- Added a new exoskeleton concept model for simulation-driven conceptual design

  of exoskeletons. The model was developed by Prof. John Rasmussen from Aalborg

  University. Please see the [web

  cast](https://www.anybodytech.com/download/simulation-driven-conceptual-design-of-exoskeletons/)

  for more details. <{ref}`See more

  <sphx_glr_auto_examples_Other_plot_ExoConcept_BoxLift.py>`>

- Added a new femoral torsion tool to apply femoral torsion to the TLEM2.0 leg

  model. This model tool was developed by Dr. Enrico De Pieri from University of

  Basel Children’s Hospital (UKBB). Please see the documentation or the [web

  cast](https://www.anybodytech.com/download/modeling-subject-specific-femoral-torsion-for-the-analysis-of-lower-limb-joint-loads/)

  by Enrico on his work and publication on femoral torsion. <{ref}`See more

  <sphx_glr_auto_examples_Orthopedics_and_rehab_plot_FemoralTorsionTool.py>`>

- Added a new knee force model example. The model shows how to calculate a

  simple estimate of the medial and lateral knee force. <{ref}`See more

  <sphx_glr_auto_examples_Orthopedics_and_rehab_plot_KneeForcesExample.py>`>

- A new box lifting motion capture model has been added. The model is based on

  data from an inertial measurement unit based suit

  ([Xsens](https://www.movella.com/products/wearables/xsens-mtw-awinda)), and

  illustrates how to connect MoCap models with objects in the environment.

  <{ref}`See more <sphx_glr_auto_examples_mocap_plot_bvh_boxlift.py>`>

**New features:**

- Many AnyScript example models now include an option to automatically create a

  video from the model. To use this functionality find and run the new

  `VideoTool.CreateVideo` operation under the Study folder. This will

  automatically run the example model, output still frames, and combine them

  into a video afterwards. The script can easily be adapted and moved to other

  examples. 

- Two new utility macros, which make it easier to create 3D grid arrays, have

  been added to the model repository. `MESHTRIPLES(xarr, yarr, zarr)` and

  `MESHGRID(xarr,yarr,zarr)`. i.e. for creating arrays of all points in a 3D

  grid array. More information in the file: {menuselection}`Body --> AAUHuman

  --> BodyModels --> GenericBodyModel --> Helper.ClassTemplates.any`

- Added a warning when the glenohumeral flexion/abduction in the mannequin

  values can cause problems as start guess for the kinematic solver. A small

  automatic pertubation of the Humerus orientation (`Axes0`) is also added in

  these cases so the shoulder rhythm will work as expected. 

- There is a new option to override more settings in the

  `DefaultMannequinDrivers` section. Now the driver type (`CType=Hard/Soft`),

  can be overridden directly by the users.

- Fixed a lack of robustness with muscle recruitment of the abdominal muscles

  (`buckle support`). Muscle recruitment could previously fail with high lumbar

  flexion. A small support "artificial muscle" has beeen added to the buckle

  segment, preventing muscle recruitment from failing. 

**Fixed:**

- Improved the selected output by adding the m. Semimembranosus and m.

  Semitendinosus contributions to the knee flexor group for both muscle activity

  (`KneeFlexorMuscleActivity`) and muscle force (`KneeFlexorMuscleForce`).

- Updated and improved origin of the hamstring muscles. The muscles biceps

  femoris, semitendinosus and semimembranosus have been updated to improve their

  moment arms in certain lunge movements. The origin points at pelvis has been

  moved down a bit to ensure the muscles wraps correctly in lunge movements,

  while maintaining the same moment arm in gait etc.

- Fixed a bug in scaling where the mass of the trunk model pelvis was used in

  some cases instead of the pelvis mass from the leg model. This could cause a

  slightly incorrect mass to be used for the pelvis segment.

- Fixed wrong sign for the WristJoint flexion variable. The values users provided in the mannequin 

  section was incorrectly interpreted as wrist extension.

- Fixed a problem where custom scaling and {bm_constant}`_SCALING_XYZ_` would 

  prevent the model from loading.

- Fixed some logical issues with the `ContactSurfaceDistanceAndVelocityDepLinPush.any` file.

   - Fix length calculation and drawing of the normal force

   - Fixed an extrapolation error when changing the internal setting

     `_SMOOTHING_FUNCTION_PROFILE_` to `FUNC_PROFILE_BSPLINE`. We now cap the

     height ratio to only be in the interpolation area. Values outside this area

     infer that no contact is present.

- Refactored the way Trunk nodes are mirrored between left and right. This is

  more consistent with the remaining body parts and handled in the cadaver data

  files.

- Fixed a problem in MoCap models where calibration studies were not run

  automatically in models that only had 3-element muscles on the upper body.

- The {ref}`example to evaluate moments arms <sphx_glr_auto_examples_Validation_plot_EvaluateMomentArms.py>` 

  now works when the shoulder rhythm is enabled. 

**Changed:**

- Improve the trunk model's strength for trunk external rotation. This change

  improves the implementation of the internal/external obliques and latissimus

  dorsi muscles. More branches are added to the internal obliques and the

  geometry is updated to better match anatomical text books. Likewise, the

  latissimus dorsi muscles is discretized into more branches so it has branches

  inserting on every vertebra.

- The `clavicle` entry in the `Anthropometric.SegmentMasses` folder have been

  renamed to `shoulder`. This makes it consistent with the rest of varaibles in

  the folder, and correctly reflect that the mass is assigned to both the

  scapula and clavicle segment. 

- The `OptimalFiberLength` and `TotalTendonLength` in the TLEM leg models are no

  longer 'DesignVar' when the parameters are also calibrated. This prevents the

  huge amount of warnings when calibrating the leg muscles.

- The `L5LContactNode`,`L4LContactNode`,`L3LContactNode`,

  `L4LContactNode`,`L1LContactNode` nodes were scaled using the Right node

  Z-axis. This is now changed to use the Left node Z-axis.

- The trunk model has been restructured in preparation for a full Thoracic

  model. This means that all the vertebra and ribs have been created in the

  model structure, but only as `AnyFolder&` references to the single rigid

  thorax segment.

  :::{note} A number of nodes have been renamed in the process. If you have

  problems loading old model due to missing nodes, please consult the new model

  examples to find the new names. 

  :::

- Updated the neutral scapula position and scapula sliding. The default neutral

  scapula position (medial rotation) has been updated to a more realistic

  position. Additionally, the node at scapula TS point, on which the scapula

  slides, has been moved to provide a more realistic clearance between the

  scapula and the ellipsoid sliding surface representing thorax.  

  Thanks to Johanna Menze (@menzejo) from the University of Bern for updating

  the model.

- Renamed a few variables named `r` which could collide with built-in `r`

  variable. 

**Removed:**

- The previously deprecated BM keyword `_LEG_MODEL_Leg_` has been completely

  removed.

- The special mannequin driver switches for the old leg model

  (`BM_MANNEQUIN_DRIVER_ANKLE_EVERSION_RIGHT/LEFT`) have been deprecated.

  Instead the switch used for the TLEM1/2 leg models

  ({bm_statement}`BM_MANNEQUIN_DRIVER_ANKLE_SUBTALAR_EVERSION_RIGHT`) can now be

  used for all leg models. 

- Renamed all deprecated ligament class names. `AnyViaPointLigament`->

  `AnyLigamentViaPoint`.

## AMMR 2.3.4 (2021-07-05)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.5060249.svg)](https://doi.org/10.5281/zenodo.5060249) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.3.4-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-3-4-8518_x64/)

**Fixed:**

- Improved the intial guess for Latissimus dorsi wrapping, so the wrapping

  solutions becomes more robust to extreeme starting postions.

- Fix a problem with `<ANYBODY_PATH_MODELUTILS>/Reactions` support files

  preventing models from loading correctly. Thanks to prof. Michael Skipper

  Andersen (Aalborg University) for finding this bug.

**Added:**

- Add small helper class template `DriverBasedOnLoadPos` which makes it easy to

  add constraints between two segments based on their load time positions.

- Refactor the way pelvis rotations are measured in the interface folder. This

  allows the user to override the default implicit global reference frame and

  thus output pelvis rotation relative to a custom frame.

- Add a new ability to evaluate trunk strength into the built-in studies for

  {ref}`joint strength evaluation

  <sphx_glr_auto_examples_Validation_plot_EvaluateJointStrength.py>`.

- New ligaments connecting the Sacrum and Pelvis segments are added. They are

  enabled togehter with the ligaments on the lumbar spine through the

  `BM_TRUNK_LUMBAR_LIGAMENTS` define statement.

**Changed:**

- The {ref}`"ADL gait (MoCap model)"

  <sphx_glr_auto_examples_MoCap_plot_ADL_Gait.py>` has been updated with most

  recent improvements from our internal projects. The marker protocol has been

  adjust and a lot of small data problems (marker dropouts etc) has been fixed.

- The posture of the {ref}`standing lift example

  <sphx_glr_auto_examples_ADLs_and_ergonomics_plot_StandingLift.py>` have been

  modified to make the example more robust.

- The `Tuber_ischiadicum` node to `Seat_contact` as it was misaligned to the

  actual bony landmark position. New `Tuber_ischiadicum` nodes have been added

  at the bony landmark. The renaming also makes it easier to identify what the

  node is used for across the ammr models.

- Changed the way the arm joint moment measures are calculated. This affect the

  output when no muscles are added to the model, and the  

  moments are now calculated and expressed in same way as when muscles are

  included. The moments are expressed in a fixed proximal reference frame

  instead of using the euler axis of rotation which may not be othogonal. All

  moment contribution also now found using search function which will be more

  robust. 

## AMMR 2.3.3 (2021-03-24)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.4616316.svg)](https://doi.org/10.5281/zenodo.4616316) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.3.3-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-3-3-8440_x64/)

**Fixed:**

  `OptimizeAnthropometricsOnOff.any` preventing models with only one leg

- Fix a problem single leg models in AnyMocap. The file

  included from being loaded.

- Removed two unused measures with references to thorax in the scapula and

  clavicula segments. The references could cause problems for ´AnyKinCOM\` which

  uses search when including segments. Thanks to Handa Kensuke from Terrabyte

  helping fix this error.

- Fix bug in initial wrapping guess for Triceps LH muscle wrapping. This bug

  could cause the left side Triceps LH muscle to wrap incorrectly at shoulder.

- Fix calibration of of element 4 and 5 of the latissimus dorsi muscle. These

  two stands of the muscles where not included correctly in the calibration

  study when using the three element muscles model.

**Changed:**

- The settings of the {ref}`Knee Simulator example <example_kneesimulator>` has been

  tweaked to make it run faster.

## AMMR 2.3.2 (2021-01-21)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.4305559.svg)](https://doi.org/10.5281/zenodo.4305559) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.3.2-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-3-2-8354_x64/)

```{rst-class} without-title

```

:::{warning} **Model results can change!**

A bug related to scaling was fixed in AMMR 2.3.2. This affect models using the

{bm_constant}`length-mass-fat<_SCALING_LENGTHMASSFAT_>` and

{bm_constant}`uniform <_SCALING_UNIFORM_>` scaling, when only Body Height is

specified. The results can change as models more accurately match the

requested body height.

:::

**Fixed:**

- Fixed tibialis anterior/posterior variables in the `SelectedOutput` section.

  The force values were represented as the `max()` of the muscle elements

  instead of the `sum()`. Thanks to Dr. Enrico De Pierie from University

  Children's Hospital Basel for fixing this issue.

- Fixed a bug when using {bm_constant}`_SCALING_LENGTHMASSFAT_` and

  {bm_constant}`_SCALING_UNIFORM_` and only specifying the body height as input.

  The segment lengths were set to slightly smaller values, which caused the

  total body height to be less than what was used as input. The bug occurred

  since a standard body height in some 'AnyMan' files was assumed to be 1.8 m

  instead of the correct 1.75 m. This fix aligns the results with

  {bm_constant}`_SCALING_XYZ_` (used in the MoCap models) and

  {bm_constant}`_SCALING_STANDARD_` (50 percentile) which were both correct.

- Fixed a problem with foot support in the {ref}`"Spine Pressure validation

  model" <sphx_glr_auto_examples_Validation_plot_WilkeSpinePressure.py>` lying

  flat on the back. The foot is now supported in the right direction.

- Fixed a problem with the Standing Model template which was using an alternate

  mode where foot constraints were disabled.

- Fixed a problem with the ADL gait MoCap example, where the batch processing

  script would report some trials as failed even if they succeeded correctly.

- Use correct times in the Xsens example MoCap (trial "S01-Trial2"). The model

  stopped walking after a number of time steps

  because `tEnd` was set after the last data point in the BVH data file.

- Fixed/improved the wrapping surface for the subscapularis muscles. The muscles

  can no longer slide off

  its wrapping surface in extreme range of motions.

- Fixed infraspinatus muscle wrapping for high arm positions. This fix adds a

  torus for infraspinatus wrapping on the humerus.

  It fixes a problem where infraspinatus could slide of its wrapping surface in

  high arm postures.

- Fix problem with latissimus dorsi muscle wrapping in over-head arm postures. A

  new wrapping ellipsoid has been added at the humeral head.

**Changed:**

- The rectus femoris muscle wrapping surface at the hip has been adjusted to

  prevent collision between the rectus femoris origin and the wrapping surface.

  Further, the surface is now used for both the straight and reflected head of

  the rectus femoris.

- Increase the discretization (StringMesh) on approx 30 muscles to improve

  wrapping.

- The results for the {ref}`"Wilke Spine Pressure"

  <sphx_glr_auto_examples_Validation_plot_WilkeSpinePressure.py>` validation

  model has been updated due to the fix for the

  {bm_constant}`_SCALING_UNIFORM_`. The updated model improved the validation

  results slightly.

- The new large scale {ref}`" MoCap model (ADL gait)"

  <sphx_glr_auto_examples_MoCap_plot_ADL_Gait.py>` has been updated so the

  results of running all standing reference trials have been added to the model.

  This means that all the dynamic trials will work even if the user forgets to

  first run the standing reference trials. It is still necessary to run the

  standing reference trials again if any changes are made to the model.

- The wrapping for the "Triceps Long Head" muscle has been updated at the

  humeral head to avoid the muscle wrapping incorrectly in over-head arm

  postures. This fix also adds a via point at the middle of the humerus to make

  the wrapping solution more robust.

- The `VideoLookAtCamera` camera class template in the AMMR now saves videos as

  they look in the model view. I.e. if things are hidden in the model view they

  will not show up in the video. This feature is enabled by a new setting

  `AnyCameraLookAt.RenderUserInterfaceViewState` in AnyBody version 7.3.2. The

  old behavior can be restored by setting

  `Camera.RenderUserInterfaceViewState=Off;` in the class template.

## AMMR 2.3.1 (2020-09-30)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.4023956.svg)](https://doi.org/10.5281/zenodo.4023956) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.3.1-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-3-1-8217_x64/)

**Added:**

- All models now run with the new wrapping solver by default, which makes all

  models with wrapping muscles run much faster. Hence the

  `AnyBody73_ExperimentalShortestPathSolver` setting is deprecated and no

  effect. The old wrapping algorithm can be enabled (not recommended) by

  setting:

  ```{code-block} AnyScriptDoc

  System.Compatibility.ShortestPathSolverVersion = 2; // 4 is default (new wrapping)

  ```

- New dedicated wrapping ellipsoid for Triceps at the glenohumeral head. This

  addition solves a few problems with overhead arm postures.

**Changed:**

- Improvements to the new {ref}`"ADL Gait"

  <sphx_glr_auto_examples_MoCap_plot_ADL_Gait.py>` example. The model now has

  muscles enabled and the marker protocol has been adjusted for a better

  shoulder posture. Also, the batch processing script now runs all trials

  without hickups.

- Improve the shoulder rhythm. Scapula position when using the shoulder rhythm

  is now determined by the Conoideum ligament driver, like how scapula kinematic

  is controlled with no shoulder rhythm. The rhythm is now controlled based on

  the **plane of elevation** and **humerus elevation** from the 'ThoraxHumerus'

  measures in the interface folder. This tracks the original specification of

  the shoulder rhythm in "De Groot, J. H. *The shoulder: a kinematic and dynamic

  analysis of motion and loading.* (1998)". One difference is still shoulder

  protraction/retraction where regression coefficients needed to be tweaked for

  the shoulder to work in the full range of motion. De Groot et al (1998) had

  only based their shoulder rhythm on arm position in front of the body (e.i.

  positive elevation planes).

- Removed metabolic power (`Pmet`) contributions from artificial muscles defined

  as `AnyMuscle` derived classes. `Pmet` contributions from tools, such as GRF

  prediction, modelled as `AnyGeneralMuscles` will no longer be part of studies'

  Pmet calculations.

- Adjusted the wrapping segments for deltoid posterior so it doesn't affect the

  muscle in overhead arm posturess.

- Teres minor now uses a torus for wrapping at the humeral head.

**Fixed:**

- Fixed incorrect values in "SelectedOutput" for four reaction moments

  variables, which were wrong due an implementation change of

  `AnyRevoluteJoint`. The following variables in

  `SelectedOutput.Trunk.JointReactionForce` were incorrect.

  - `C2C1FlexionExtensionMoment`

  - `C2C1LateralMoment`

  - `C1C0AxialMoment`

  - `C1C0LateralMoment`

  The values were wrong because they referered to the internal generalized force

  output of `AnyRevoluteJoint`.  (i.e. `Constraints.Reactions.Fout`). The values

  `Fout[3]` and `Fout[4]` can not be interpreted as real physical reaction

  moments because `AnyRevoluteJoint` is implemented using euler angles.

  :::{warning}

  The use of generalized forces (`Fout`) is

  discouraged, and will be removed in the AMMR. Instead use the force and moment

  variables in `AnyRefFrameOutput` subfolder instead.

  :::

- Fixed incorrect volume of Satorius muscle in TLEM1 and TLEM2. Thanks to Dr.

  Adam D. Sylvester from Johns Hopkins School of Medicine and Dr. Patricia A.

  Kramer from the University of Washington for pointing out the error. The error

  occurred since the satorius in the original [TLEM1

  paper](https://doi.org/10.1016/j.clinbiomech.2006.10.003) was muscles in

  series with a pseudo insertion/origin on the femur. Both of these muscle

  elements were therefore listed with the full PSCA of Satorius. This detail was

  missed in the AnyBody TLEM1 and TLEM2 implementation where Satorius was

  implemented two parallel elements (TLEM1) and a single element (TLEM2). Hence

  the Satorius had twice the correct volume.

- Fixed a problem when C3D files are malformed and doesn't contain a

  `FORCE_PLATFORM.ZERO` variable. The C3D standard specifies that the variable

  should exist but since it not used in the models the reference could just be

  remvoed.

- Fixed an problem which prevented the ANSUR scaling plugin example from working

  correctly 7.3.0

- Fixed incorrect output of `ThoraxHumerus` interface measures. The mesure

  wasn't used by other parts of the AMMR until now so the bug had no

  consequences in previous AMMR versions.

- Fixed mass calculation of the foot and talus segments in the `GaitVaughan`

  validation model.

**Removed:**

- Removed unused `BM_TRUNK_CERVICAL_LIGAMENTS` switch which had no effect since

  there were no ligaments defined in the neck model.

## AMMR 2.3 (2020-07-07)

[![Zenodo link](https://zenodo.org/badge/DOI/10.5281/zenodo.3932764.svg)](https://doi.org/10.5281/zenodo.3932764) [![AnyBody link](https://img.shields.io/badge/Included_with_AnyBody-7.3.0-yellowgreen)](https://www.anybodytech.com/download/anybodysetup-7-3-0-8065_x64/)

**Added:**

- Added a new Full Body {ref}`"ADL Gait"

  <sphx_glr_auto_examples_MoCap_plot_ADL_Gait.py>` MoCap Example based on the

  "Rehazenter Adult Walking Dataset" by [Schreiber and Moissenet

  (2019)](https://doi.org/10.1038/s41597-019-0124-4). The model is configured to

  run all 50 subjects and 1145 trials in the data set. However, you must

  download the actual data separately from

  [FigShare](https://figshare.com/articles/dataset/A_multimodal_dataset_of_human_gait_at_different_walking_speeds/7734767)

  where it is hosted under a Creative Commons license.

- New GUI plugin which uses statistical information from the ANSUR database to

  set anthropometric values in AnyBody. A small example model is added

  `Examples/StatisticalScalingPlugin` which shows how to use the plugin.

- Added a new option to optimize "head width" / "head depth" / "neck length" /

  "Tibia torsion" / "hand breadth" as part of the `OptimizeAnthropometricsOnOff`

  class_template in the AnyMoCap framework.

- Added new rotation vector based measures in the interface folder for the arm

  degrees of freedom. These measures (`Interface.Right.RotVectorMeasures.*`) for

  joint angles do not have a clinical meaning but are useful when exporting and

  importing joint angles.

- Added a new {ammr:bm_statement}`option for setting custom muscle calibration

  <BM_CALIBRATION_TYPE>`. I.e. an option which disables calibration so the user

  can add their own code.

  ```{code-block} AnyScriptDoc

  #define BM_CALIBRATION_TYPE _CALIBRATION_TYPE_CUSTOM_

  #include "<ANYBODY_PATH_BODY>\HumanModel.any"

  // Manually include calibration code.

  Main.HumanModel.Calibration = {

    #include "<ANYBODY_PATH_BODY>/LegTLEM/Calibration/CalibrationSequenceRight.any"

    #include "<ANYBODY_PATH_BODY>/LegTLEM/Calibration/CalibrationSequenceLeft.any"

    #include "<ANYBODY_PATH_BODY>/Arm/Calibration/CalibrationSequenceRight.any"

    #include "<ANYBODY_PATH_BODY>/Arm/Calibration/CalibrationSequenceLeft.any"

  };

  ```

  :::{note}

  If you need to modify the calibration code, copy it to the application folder 

  and include it from there.

  :::

**Changed:**

- All application examples use the new and fast experimental wrapping algorithmn

  in the AnyBody Modeling System 7.3.0. A notice is shown in the output window to

  remind the users that the algorithm is enabled. The experimental wrapping

  algorithm

  can be used in all models by setting:

  ```{code-block} AnyScriptDoc

  System.Compatibility.AnyBody73_ExperimentalShortestPathSolver = On;

  ```

  :::{seealso}

  The release notes for the AnyBoy Modeling System has more information

  on the new and faster experimental wrapping algorithm.

  :::

- Split the SternoCleidomastoid muscle up into two muscle branches. This makes

  the two heads of the muscle able to work independent.

- Add extra muscle-via points on for the Latissimus dorsi on the Thorax segment.

  This change fixes wrapping problems where the Latissimus dorsi muscle strands

  would slide off the wrapping surface representing the thorax segment.

- Renamed all deprecated muscle class names. `AnyViaPointMuscle`-> `AnyMuscleViaPoint`

  and `AnyShortestPathMuscle`-> `AnyMuscleShortestPath`

- The AnyMoCap framework now uses a more robust way of saving optimized

  parameters from the parameter optimization studies, when using AMS 7.3.

- Pectoralis Major has been split from 5 into 10 branches to better represent

  the real fiber direction of the muscle. Similarly the origins of the new

  Pectoralis Major muscles have been moved to align with real anatomy of the

  muscle.

- Changed the format used by the AnyMoCap models when transferring joint angles

  from the mrker tracking step to inverse dynamics simulation. The joint-angles

  for the arms and pelvis rotation are now written in terms of rotation vectors

  instead of Euler angles. This fixes potential singularity problems. The

  temporary joint angle files produced by the MoCap models have been renamed to

  indicate the data is no longer Euler angles.

  :::{note}

  This means that the data in these temporary files can not easily be interpreted from a clinical

  point of view. Create your own output files with physiological joint angles if you need to export that data.

  :::

- The body model configuration plugin have been updated. This fixes a number of

  smaller issues and bugs.

- The via points of PeroneusLongus and PeroneusBrevis have been adjusted to

  ensure the muscles maintain moment arm in a larger range of motion.

- Rectus abdominis muscle has been split into eight branches to better represent

  geometry, than what is possible with one branch. The strength between the new

  branches has been split evenly. Additionally, the PCSA has been increased from

  $2.6 cm^2$ to $7.9 cm^2$ according to McGill et al 1988, the $2.6 cm^2$

  originates from a study with elderly subjects, the McGill study used mid age

  subjects.

- Adjusted the origin of Serratus Anterior 1 so it aligns with the geometry of

  the rib on which it originates.

- Changed the way the deltoid wrapping surfaces move relative to humerus. This

  changes makes it wrapping more robust to deltoid flipping over to the wrong

  side of the wrapping surface in extreme postures.

- The BVH virtual marker protocol have been adjusted. If older BVH models is

  used with AMMR 2.3 a warning/error is triggered to warn the user that the

  protocol needs to be updated.

**Fixed:**

- Fixed problems with the following variables in the `SelectedOuput` folder:

  `HipAbductorMuscleForce`, `HipExtensorMuscleActivity`,

  `HipExtensorMuscleForce`, `KneeFlexorMuscleActivity` and

  `KneeFlexorMuscleForce`.

- Markers on BVH stick-figure model in the AnyMoCap framework now respects the

  setting: `Main.ModelSetup.BVHFileData.ModelDrawOnOff`.

- Fixed occasional kinematics problems with FullBody MoCap models during inverse

  dynamics simulation. This was fixed by changing the internal representation of

  the shoulder joint angles when transferring kinematics from marker tracking to

  the Inverse Dynamics simulations.

- Fixed a problem with the Body Model configurator plugin showing reporting: "No

  Human Model" when a configuration file already exists.

- Corrected the location of via points of Flexor_Digitorum_Superficialis_Digit2

  to avoid wrapping surface collisions in certain hand positions.

- Corrected the location of via points of Flexor_Digitorum_Superficialis_Digit3

  and 4, Palmaris_Longus, Extensor_Carpi_Radialis_Brevis and

  Extensor_Digitorum_Digit2 to avoid wrapping surface collisions in certain hand

  positions.

- The muscle via point representing the bicipital groove have been adjusted to

  ensure the Biceps Caput Longum wraps the humeral head correctly in

- Fixed a bug in the FreePosture model, where input for the Left/Right arm

  drivers was switched in some places.

- Fixed missing visual color indication of force plate forces for type 3 force

  plates in the MoCap models.

- Fixed a bug with in the built-in studies to {ref}`evaluate arm joint strength

  <sphx_glr_auto_examples_Validation_plot_EvaluateJointStrength.py>`. The range

  of motion for the left arm elbow pronation/supination were not correct.

- Fixed a bug preventing the model from loading with with

  {bm_constant}`_SCALING_XYZ_` and both legs excluded.

- Fixed the load-time position of the head segment when neck scaling is changed.

- Fixed a problem with visualization in GRF prediction class which prevented GRF

  prediction to be used with moving base frames (i.e. treadmills). Thanks to

  Marco Antonio Marra (University of Twente) for fixing this.

- Fixed the marker node orientation for BVH (inertial based) MoCap models in the

  CreateMarkerDriver class template. Marker orientations are now handles

  identical to normal C3D based markers.\`\`\`

- Fixed incorrect wrapping of teres minor and teres major at extreme shoulder

  range of motion. New wrapping surfaces have been added. One in Humerus and one

  on scapula. The meet at the glenohumeral joint to prevent the muscles from

  sliding to the wrong side of the humerus.