Modeling and Simulation Roadmap

The worldwide team of developers for OpenSim, Moco, OpenCap, AddBiomechanics, OpenSense, OpenSim Creator, and other related software, models, and datasets has an ambitious set of plans for 2022 and beyond. This roadmap is based on the valuable input from the worldwide community that is using these resources.

Modeling and Simulation Resources

There is a growing family of desktop and web applications, models, datasets, and training materials to support modeling and simulation of movement. Some of these resources are developed at Stanford, but many are also developed and shared by our wider developer and user community. Resources include:

• OpenSim  for musculoskeletal modeling and simulation, including the Desktop Application and API
  
• OpenSim Moco, an optimal control toolkit that is distributed with OpenSim
• OpenSense for IMU-based biomechanical analysis, also distributed with OpenSim
• OpenCap  for video-based movement analysis and dynamic simulation
• Sit2Stand  for assessment of sit to stand biomechanics via single video
• addBiomechanics a web application for automatic scaling and inverse kinematics, as well as integrated data sharing
• OpenSim Creator for building models
• Scone for predictive musculoskeletal simulation
• Simtk.org, a general purpose website for sharing biomechanics and other biomedical software, models, data, and other resources
• Musculoskeletal models, some of which are listed on the musculoskeletal model library, with many others shared on Simtk.org
• Motion capture data and simulations as shared on Simtk.org
• Scripts and plugins for added utility and functionality. These are distributed with OpenSim and also available on Simtk.org and GitHub.
• Tutorials, videos, and other documentation, which you can find on this online wiki documentation site
• A User Forum for OpenSim.

Our Plans

We have plans to expand and improve on these shared resources. Our priorities include:

• Liberate siloed data and foster data sharing to improve reproducibility and usher in the next generation of biomechanics tools that transform human health and performance
  • AddBiomechanics, which provides automated scaling and inverse kinematics, is making it easier to process and share large amounts of motion capture data. This will create a large database of movement data for the whole community. One development we are working on next is supporting dynamics, in addition to kinematics. 
    
  • OpenCap provides a common platform for estimating kinematics and kinetics from video capture with iOS or other devices. We will continue to expand and test this software and welcome use and contributions from the community. Some specific plans include enabling simulations to be run in the cloud and enabling easy plotting of kinematic and kinetic results.
    
• Create fast and accurate software, models, and simulations for studying the biomechanics of movement
  • Model building is usually cumbersome and time-consuming. OpenSim Creator is an easy-to-use GUI to enable scientists to efficiently create and validate new OpenSim models. A team at TU Delft is building this tool and will continue to improve usability and add modeling features.
    
  • Muscle wrapping and path calculations are limiting factors for simulation speed. We are working to update the algorithms used to calculate muscle-tendon lengths around wrapping surfaces, adding more tests to ensure accuracy and robustness, and also exploring function- or ML-based representations of muscle quantities like moment arms.
  • Contact modeling is another common bottleneck for simulation speed. Members of the community are exploring spring-based compliant contact. We also plan to explore whether machine-learning, data-driven methods can provide accurate and fast contact force estimations. 
  • Optimal control methods, such as direct collocation, are enabling researchers to efficiently solve more complex simulations than ever before. We will continue to improve the speed of our algorithms in Moco. We plan to add support for Automatic Differentiation in OpenSim and leverage function-based representations of muscle paths with OpenSim Moco.
  • Incorporating medical imaging can aid model customization. The community is exploring means to efficiently use these data to improve the accuracy of models and simulations.
• Move biomechanical analysis outside of the lab and reduce resource barriers like cost, time, and space by leveraging wearable sensor and video data, along with a combined ML and biomechanics-based approach to increase automation, accuracy, and speed
  • Tools like OpenCap and AddBiomechanics are helping us to achieve these goal.
  • OpenSense allows assessing biomechanics with wearable sensors. We will continue to test this tool in new application areas (e.g., the upper extremity) and work on ways to make the tool easier to use (e.g., reducing the number of sensors required to achieve high-quality results).
  • Web-based interfaces to our software will allow anyone to use OpenSim without a burdensome install process. We will work on making OpenSim accessible via a web app, which will also support tighter integration with other web-based tools like OpenCap and AddBiomechanics
  • The community is also focusing on translational impact of modeling and simulation by, for example, testing the ability of simulation tools to improve screening or diagnosis and predict changes due to a device or treatment.  
• Create and share beautiful visualizations of movement to educate, inform, and inspire
  • We are planning to create improved web-based visualization and explore Blender integration for rendering
• Share teaching materials that help accelerate research and train the next generation of scientists to apply biomechanics to make new discoveries and have translational impact.
  • We are creating highly-accessible examples for teaching through tools like OpenCap.
  • The recent Biomechanics of Movement textbook and associated resources is a valuable resource for the community and we will continue to improve integration, for example with additional modeling-based examples and tutorials.
• Empower scientists and developers from around the world to contribute to these goals and achieve their own research or translational goals
  • We want to continue to engage and support developers with improved documentation and a streamlined build process.
  • Python and Matlab scripting allows scientists without extensive development experience to push the bounds of what is possible with simulation. We are making these tools more accessible by providing resources like a conda package and example colab/Jupyter notebooks. 

We're planning to roll out these improvements over a series of software and resource releases. For developers, if you're interested to see what we're working on right now, or would like to suggest a new feature, visit the GitHub page, where you can also download the very latest version of OpenSim and contribute your own code to the software.

We also welcome feedback and additional ideas to improve this set of resources. Please send comments via email to opensim@stanford.edu.