Modelling the human-exoskeleton interaction interface - Benchmarking MuJoCo's deformable model

TL;DR The authors benchmark MuJoCo's flexcomp soft-body models against a COMSOL finite-element model on indenting and compressing silicone tissue, including a simplified thigh, as groundwork for simulating the human-exoskeleton interface. MuJoCo follows the force trends (correlation 0.97-0.99) but is stiffer, captures only 4-24% of the finite-element hysteresis, and is faster only in the sphere-indentation case.
What they do
An exoskeleton pushes on the body through straps and cuffs over skin, fat and muscle, and that soft tissue absorbs and releases part of the assistance. MuJoCo can simulate soft bodies with its flexcomp library in the same scene as musculoskeletal and exoskeleton models, but its accuracy for soft tissue had not been checked against finite elements. The authors benchmark flexcomp against COMSOL in three simulated cases of increasing complexity, as a first step toward modelling the human-exoskeleton interface.
How it works
- Cases. (1) A 10 mm-radius sphere indents a 70x70x25 mm block by 9 mm; (2) a rigid plate compresses the same block by 5 mm; (3) a plate compresses a cylindrical soft-tissue thigh, bonded to a rigid femur, by 10 mm. Loading speeds are 1-5 mm/s.
- Material. The tissue is a homogeneous, nearly incompressible silicone (Ecoflex 00-10) chosen to mimic the compressive response of human thigh tissue.
- MuJoCo. Linear tetrahedral flexcomp with the Saint Venant-Kirchhoff (StVK) or the experimental Stable Neo-Hookean (SNH) model, Young’s modulus 40 kPa, Poisson’s ratio 0.485, Rayleigh damping and a 2 ms time step.
- COMSOL reference. Quadratic tetrahedra with a three-term Yeoh hyperelastic model, a two-branch viscoelastic model and a mixed displacement-pressure formulation. The comparison covers force and strain energy against depth, hysteresis and wall-clock time.
Results
- Force-indentation curves correlate with COMSOL at 0.97-0.99 in every case, but MuJoCo is stiffer. RMSE is 0.8 N (StVK) and 3.6 N (SNH) for the sphere, 1.8 N and 13 N for the block compression, and 39.4 N and 53.6 N for the thigh.
- Hysteresis is badly underestimated: in case 1, 1.81-3.73% (StVK) and 0.30-1.49% (SNH) against 7.67-15.90% for COMSOL. Across cases StVK captures 12-24% of COMSOL’s hysteresis and SNH 4-11%.
- MuJoCo is 2-2.5x faster than COMSOL for the sphere, but about 3x slower for block compression and about 16x slower for the thigh, where many discrete contacts enlarge the constraint problem.
Limitations
The authors conclude that flexcomp did not reproduce COMSOL’s soft-tissue behaviour and is not a valid substitute in these cases. They attribute the extra stiffness to volumetric locking of linear tetrahedra and the missing hysteresis to Rayleigh damping standing in for a viscoelastic model; contact forces also oscillate in cases 2 and 3. Both sides are simulations with literature parameters and different constitutive models and meshes, the geometries are simplified, and no measured tissue data enter the comparison.
Figures




Abstract
The human-exoskeleton interaction interface, consisting of the exoskeleton attachment and the underlying human soft tissue, can affect the transfer of assistance. MuJoCo's \textit{flexcomp} has been previously used to model the soft tissue, but its accuracy has not been benchmarked against a finite element model. We benchmarked the \textit{flexcomp} Saint Venant-Kirchhoff (StVK) and Stable Neo-Hookean (SNH) models against COMSOL's Yeoh model in three cases: \textit{case 1}) spherical indentation, \textit{case 2}) compression, and \textit{case 3}) compression of a thigh. We compared force and strain energy as a function of indentation depth, hysteresis and computation time. MuJoCo's models capture force trends with a correlation of 0.97-0.99, but showed a stiffer response, which could be attributed to the volumetric locking of the linear finite elements. The MuJoCo models systematically underestimated hysteresis as MuJoCo's Rayleigh damping could not substitute a viscoelastic model. Moreover, MuJoCo's computational advantage was speed dependent: it was 2-2.5 times faster for spherical indentation, but COMSOL was faster for the other cases, which have more degrees of freedom and contacts. Although MuJoCo's \textit{flexcomp} offers a combination of soft-tissue modelling and multibody dynamics, the limitations regarding the stiffening and hysteresis should be addressed to be used for the analysis of the human-exoskeleton interaction interface.
Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.
