Tweet
opty 1.4.0 is here!. Here is an optimal solution of a person "running" on the moon:
human-gait-moon-3mps.gif
What can it do?
- identify the closed loop human control system from measurement data (motion, forces, EMG, etc.) [standing balance example]
- identify the open loop human control inputs from measurement data (similar to inverse dynamics without necessarily needing force measurements) [gait example]
- find optimal open loop controls used to move a person (highest jump, fastest race, maximum force, etc.) [cycling race example, sit-to-stand example]
- identify active or passive assistance device parameters for optimal motion or from motion tracking
Why use it?
- It is very fast, solves most problems in a few minutes by generating very efficient parallelized C code.
- Handles 3D dynamics models with < ~40 rigid bodies and many millions of operations in the equations of motion.
- Accessible, written all in Python and now has 20+ examples, including some biomechanics ones.
- Inputs can be joint torques or muscle activation in biomechanics.
- Runs on Mac, Linux, and Windows.
- It is free and open source.
Why not use it?
- You have to write Python to make it run.
- It is free and open source.
- You have to be able to write the differential algebraic equations of your system (or generate them).
- Models must be defined symbolically.
- It doesn't use the fanciest collocation methods (mesh refinement or higher order polynomial methods), but I haven't found that yet to be any issue.
- It does not have a rich set of pre-made biomechanics models.
- It is a local optimizer, so some art is required to get a global minima.
- Currently only one muscle model (De Groote 2016).
- No fancy graphics.
- Something else to learn.
Thanks to the help of Sam Brockie, Timo Stienstra, Christoph Konrad, Peter Stahlecker, and the SymPy Community for this release!
Direct questions, bug reports, and feature requests to the issue tracker on Github: https://github.com/csu-hmc/opty
Thanks,
Jason K. Moore
TU Delft
human-gait-moon-3mps.gif
What can it do?
- identify the closed loop human control system from measurement data (motion, forces, EMG, etc.) [standing balance example]
- identify the open loop human control inputs from measurement data (similar to inverse dynamics without necessarily needing force measurements) [gait example]
- find optimal open loop controls used to move a person (highest jump, fastest race, maximum force, etc.) [cycling race example, sit-to-stand example]
- identify active or passive assistance device parameters for optimal motion or from motion tracking
Why use it?
- It is very fast, solves most problems in a few minutes by generating very efficient parallelized C code.
- Handles 3D dynamics models with < ~40 rigid bodies and many millions of operations in the equations of motion.
- Accessible, written all in Python and now has 20+ examples, including some biomechanics ones.
- Inputs can be joint torques or muscle activation in biomechanics.
- Runs on Mac, Linux, and Windows.
- It is free and open source.
Why not use it?
- You have to write Python to make it run.
- It is free and open source.
- You have to be able to write the differential algebraic equations of your system (or generate them).
- Models must be defined symbolically.
- It doesn't use the fanciest collocation methods (mesh refinement or higher order polynomial methods), but I haven't found that yet to be any issue.
- It does not have a rich set of pre-made biomechanics models.
- It is a local optimizer, so some art is required to get a global minima.
- Currently only one muscle model (De Groote 2016).
- No fancy graphics.
- Something else to learn.
Thanks to the help of Sam Brockie, Timo Stienstra, Christoph Konrad, Peter Stahlecker, and the SymPy Community for this release!
Direct questions, bug reports, and feature requests to the issue tracker on Github: https://github.com/csu-hmc/opty
Thanks,
Jason K. Moore
TU Delft