v1.3.2

Added

• Treatment Optimization plotting function plotTreatmentOptimizationResultsFromSettingsFile() can optionally take an override results directory as a second argument to plot results from a directory other than the one in the given settings file

Fixed

• Neural Control Personalization plots requiring multiple plot windows will correctly format overflow windows to match the first
• Prescribed coordinates in Design Optimization are saved to results correctly for free final time problems

Changed

• Updated calls to the MATLAB size() function to prevent compatibility warning messages on MATLAB R2024a and newer

v1.3.1

Added

• Marker tracking in Treatment Optimization (TO) can be used to track a subset of directions in the ground frame (default x, y, and z directions)
  • To change the included subset, add <axes>true true true</axes> with the first, second, or third entry set to false for an excluded direction
• TO synergy vectors can be used and saved normalized by either sum or magnitude of a given value (default sum to 1)
  • Inside the <RCNLSynergyController>, add <synergy_vector_normalization_method> set to sum or magnitude and add <synergy_vector_normalization_value> set to the desired vector sum or magnitude
• The root directory now includes installation instructions for the pipeline (How to Install.pdf)

Fixed

• TO runs using point kinematics will no longer crash with OpenSim 4.5.1 linked to MATLAB
• TO marker tracking will track multiple markers correctly at the same time

Changed

• Optimizing synergy vectors during TO is significantly faster
• TO plotting functions will automatically convert rotational input quantities from degrees when comparing to radians

v1.3.0

Added

• Joint Model Personalization can scale point constraints with bodies
• Individual Muscle-Tendon Personalization (MTP) design variables can be included or excluded
• MTP supports a series of <MTPTask>s, setting up multiple-stage optimizations with individual sets of design variables and cost terms
• Plotting function plotMtpHillTypeMuscleParamsCompare() can compare sets of muscle parameters optimized by MTP
• Ground Contact Personalization (GCP) can optimize a change in experimental electrical center values to correct for force plate errors and produce ground reaction moments more consistent with physical models
  • Individual electrical center shift dimensions can be added as design variables <electricalCenterX>, <electricalCenterY>, and <electricalCenterZ>
  • Electrical center adjustment can be regularized with the electrical_center_shift cost term
  • An updated version of the experimental ground reactions file will be saved with the new electrical center
• GCP saves updated full-body kinematics and modeled ground reactions with modified foot motion if all <GCPContactSurface>s have the same time range
• GCP kinematic periodicity relative to experimental motion can be enforced with the new kinematic_periodicity cost term
• Model personalization plot windows have titles indicating the data displayed
• Treatment Optimization (TO) now uses a tracked_quantities_directory for experimental data or data to track and an initial_guess_directory for initial values
  • Missing intial guess values will be copied from tracked quantities, and missing quantities to track will be copied from initial guess data if available
• TO states and controls can optionally have minimum search bounds set in addition to search scale factors
  • Fields ending in scale_factor have an equivalent minimum_range
• TO cost calculations can be normalized within individual cost term types instead of by the total number of cost terms by setting the optional <normalize_cost_by_term_type> to true
• New TO cost terms:
  • Controller shape tracking
  • Controller minimization
  • Scaled controller tracking
    • Adding a <scale_factor> to a controller_tracking cost term will track a scaled version of the reference control
• User-defined TO model functions can be used to modify either an element of the OpenSim model or part of the states or controls
  • This allows users to create their own rules for modifying states or controls with custom parameters or feedback mechanisms before model dynamics are calculated
• TO joint velocities can be plotted with the plotTreatmentOptimizationJointVelocities() function
• New plotting functions for all tools will plot relevant results from settings files when run in the same directory as the settings file. Functions include:
  • plotJmpResultsFromSettingsFile()
  • plotMtpResultsFromSettingsFile()
  • plotNcpResultsFromSettingsFile()
  • plotGcpResultsFromSettingsFile()
  • plotTreatmentOptimizationResultsFromSettingsFile()
• Using the OpenSim 4.5.1 API linked with MATLAB is now supported

Fixed

• MTP plots can properly display absolute length adjustments from Muscle-Tendon Length Initialization
• GCP updates tracked experimental ground reaction moments to use the modified spring resting length for consistency

Changed

• GCP rotation tracking allowable error should now be given in radians to be consistent with TO cost terms
• Plot visuals have been improved and made more consistent between tools with new MATLAB grid features
• TO computes metabolic cost with a MATLAB implementation of a Bhargava metabolic cost model

v1.2.0

Added

• MTP can now use translational coordinates spanned by muscles.
• Added a plotting function to show the distribution of spring stiffnesses in GCP.

Fixed

• Fixed a bug where Treatment Optimization tracking terms were using the wrong time array
• MTP now properly saves passive model moments when input passive moment data has multiple columns with non-zero data.
• Fixed a bug with plotting JMP results where the marker names ordering was not respected.
• Fixed a bug with GCP damping terms not being applied correctly.
• Fixed a bug where the wrong muscle tendon length file was parsed in Treatment Optimization.
• Fixed a bug with the incorrect time array being used for the initial guess and dependency finding steps of Treatment Optimization.

Changed

• Treatment Optimization now no longer splines results back to the initial time points, instead using the collocation time points as the final results. This reduces numerical inconsistencies between the tracking and verification steps.
• Some Treatment Optimization plotting functions have been improved.
• The surrogate model has been sped up.
• The surrogate model now uses an independent data directory. A script has been added (src/SurrogateModelCreation/surrogateKinematicsScript.m) to generate Latin hypercube sample (LHS) kinematics for the surrogate model. This improves accuracy when finding novel motion.
• .sto file parsing has been sped up.
• Improved the quality of the Treatment Optimization initial guess.
• Changed the way Treatment Optimization results are saved to make VO easier.

v1.1.0

v1.1.0

Added

• Joint Model Personalization (JMP) tasks can contain a <marker_list> explicitly naming markers to track
• Plotting functions for Muscle Tendon Personalization (MTP)
• Plotting function for overall variability accounted for (VAF) and activation RMS error for Neural Control Personalization (NCP)
• Treatment Optimization can use both torque and synergy controllers, as well as both types in the same Model
• Treatment Optimization can modify a subset of model coordinates, taking prescribed values from input data for others
• Synergy-driven Treatment Optimization can optionally adjust synergy vectors
• Synergy-driven Treatment Optimization tools automatically create surrogate polynomial muscle models instead of requiring a separate tool
• Design Optimization can solve open/free final time problems
• New Treatment Optimization cost terms:
  • Marker position tracking
  • Joint power minimization
  • Joint energy generation goal
  • Joint energy absorption goal
  • Propulsive impulse goal
  • Braking impulse goal
  • Muscle activation minimization
  • External torque control minimization
  • Whole body angular momentum minimization
  • Relative walking speed goal
  • Relative metabolic cost per time
  • Relative metabolic cost per distance traveled
• New Treatment Optimization constraint terms:
  • Initial state position
  • Limit muscle activation
  • Limit normalized fiber Length
  • Synergy weight sum
• Plotting functions for Treatment Optimization results

Fixed

• JMP preserves all joint parameters through multiple tasks
• Muscle-Tendon Length Initialization tracks passive data with extra muscle analysis files present
• NCP accurately tracks joint moments with multiple synergy groups

Changed

• Ground Contact Personalization (GCP) defines a contact surfaces with a <hindfoot_body> instead of a <toes_coordinate> or <toes_joint> in XML
• Treatment Optimization (Tracking, Verification, and Design Optimization) settings have changed significantly due to changes to how states and controls are handled
• Treatment Optimization uses GCV splines to fit data instead of B-splines
• Treatment Optimization includes coordinate accelerations in controls instead of coordinate jerks