Bioptim, a Python framework for Musculoskeletal Optimal Control in Biomechanics

Abstract

Musculoskeletal simulations are useful in biomechanics to investigate the causes of movement disorder, to estimate non-measurable physiological quantities or to study the optimality of human movement. We introduce bioptim, an easy-to-use Python framework for biomechanical optimal control, handling musculoskeletal models. Relying on algorithmic differentiation and the multiple shooting formulation, bioptim interfaces nonlinear solvers to quickly provide dynamically consistent optimal solutions. The software is both computationally efficient (C++ core) and easily customizable, thanks to its Python interface. It allows to quickly define a variety of biomechanical problems such as motion tracking/prediction, muscle-driven simulations, parameters optimization, multiphase problems, etc. It is also intended for real-time applications such as moving horizon estimation and model predictive control. Six contrasting examples are presented, comprising various models, dynamics, objective functions and constraints. They include data-driven simulations (i.e., a multiphase muscle driven gait cycle and an upper-limb real-time moving horizon estimation of muscle forces) and predictive simulations (i.e., a muscle-driven pointing task, a twisting somersault with a quaternion-based model, a position controller using external forces, and a multiphase torque-driven maximum-height jump motion).

Can lower-limb exoskeletons support sit-to-stand motions in frail elderly without crutches? A study combining optimal control and motion capture

“…One can use the Rigid Body Dynamics Library (RBDL; Felis, 2017) with MUSCOD-II (Leineweber et al, 2003) to solve multi-phase OCPs. Another available library that can solve multi-phase OCPs with the direct multiple-shooting method is bioptim (Michaud et al, 2023). It is open-source and utilizes biorbd (Michaud and Begon, 2021) and CasADi (Andersson et al, 2019) to perform optimization problems through a Python interface.…”

Moving Horizon Estimation of Human Kinematics and Muscle Forces

“…Results are shown in Section III-A. The problem was implemented using Bioptim [28], a Python library dedicated to solving biomechanical OCP, leveraging the MHE preimplemented class. Non-linear programs were solved using Acados [29], with its real-time iteration scheme [30] to improve the robustness and computational cost of the convergence of the subproblems.…”

“…In [17], we proposed a real-time muscle force estimation algorithm based on: optimal control problem (OCP), forward dynamics, moving horizon estimator (MHE), and simulated data. An originally intractable OCP was split into a series of smaller subproblems solved at high frequency (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30). We showed that forward methods are suitable for real-time applications, providing muscle forces and joint kinematics satisfying motion dynamics.…”

Optimization-Based Motion Generation for Buzzwire Tasks With the REEM-C Humanoid Robot

“…We used the CasADi framework (Andersson et al, 2019) to setup the NLP, along with the Interior Point Optimizer (Wächter and Biegler, 2006) using the MUltifrontal Massively Parallel sparse direct Solver (Amestoy et al, 2019(Amestoy et al, , 2001. The Rigid Body Dynamics Library (Felis, 2017) was used to compute the forward kinematics (κ) and the inverse dynamics (τ), using RBDL-Casadi bindings (Michaud et al, 2021) to ensure gradients could be calculated through the kinematics functions for the NLP solver.…”