Postdoctoral position - Modeling and simulating human biomechanical variability to co-optimize exoskeleton architecture, actuation, and assistance

Inria Talence Talence, France

About the position

Context

This postdoctoral position will take place within the Auctus team at the Inria center of the University of Bordeaux (Talence). The Auctus team aims at meeting challenges of collaborative robotics for humans at work. The team's research is divided into three scientific axes: analysis and modeling of human behavior (biomechanical and cognitive); human-robot interaction and coupling; design and control of cobotic systems. This project falls at the intersection of the first and third axis, by participating in the definition of a human-centered co-design framework for assistive device, specifically exoskeleton. The project will benefit from the hardware and software resources of the lab experimental platform.
External collaboration w ith the Rainbow team at Inria Rennes are planed to evaluate the co-design approach on the optimization of their upper-limb exoskeleton prototype.

Assignment

Exoskeletons are a particularly promising assistive solution for human at works as they provide motion-force support through direct interaction with the operator [
DeLooze16
, Voilqué19]. But, the complete coupling between the human and assistive device can constrain rather than help if the exoskeleton is not perfectly suited to the user, as often described by a misalignment of the exoskeleton’s joint instantaneous center of rotation with respect to the human kinematics [
Näf18
]. Such a kinematic misalignment induces unwanted forces and torques on the user that can result in torsional shear or soft tissue compression [Schiele08, Jarrassé11].
Moreover, human individuals exhibit large morphological variability [Loper15] - such as differences in body shape, mass distribution, and muscular capacity - which can lead to different motion even when performing a same given task.
Although adaptive control strategies can adjust the exoskeleton assistance to different user capacity [Gallois25] or motion strategy [Souza24], it cannot compensate for kinematic incompatibility or for misplaced attachment points resulting from a rigid exoskeleton design mounted on users with diverse morphologies. Suitability of such assistive devices for individual users therefore requires adaptive design solutions, for examples flexible, reconfigurable, or redundant mechanisms, that explicitly account for the specificity of human biomechanical variability. This postdoctoral project aims to contribute to a user-centered exoskeleton co-design methodology that integrate human biomechanical models within the design optimization
.

Main activities

T o adapt designs of exoskeletons to diverse users, one should first better understand variability in human morphological and musculoskeletal models, and then integrate human variability within the co-optimization process. This postdoctoral project will contribute to existing simulation-driven co-design frameworks by incorporating human biomechanical models in the simulation that drives the optimization of both the mechatronic architecture and controlled assistance of the system.
The project will specifically focus on upper-limb models and assistive exoskeletons. Several axes of research will be covered by the candidate:
-
Modeling human biomechanical variability through statistical population models
:
Existing PCA-based statistical models of human variability, such as SMPL-X [Pavlakos19], will be reviewed to identify principle axes of variation of both the human morphology and kinematic chain. Design guidelines on the needed adaptability of the exoskeleton should be derived from this motion-force variability analysis. A set of representative human envelops and associated musculoskeletal models
[K eller23] will be generated for a given population, into a P inocchio-modeling [Carpentier19] compatible format, to embed human variability within the simulation that will later drive exoskeleton design optimization
.
-
Optimizing the exoskeleton design through a co-design framework
: Adapting exoskeleton design to user variability can be done by adding redundant or passive degrees of freedom to the kinematic chain [Manzano25], by designing self-alignment mechanisms [Cempini12, Schorsch14], or by integrat ing compliance through flexible mechanisms at the anchor point or within the mechanical structure [Lee16]. These adaptation solutions will be studied and compared in terms of possible adjustment levels: degrees of freedom, direction and range of adjustment, etc. A set of design parameters will be defined for the exoskeleton for the architecture, actuation, and mechanism parameters. We will develop a global simulation-driven co-design framework to optimize this set of parameters jointly with the exoskeleton assistance. The optimization constraints and costs will be formulated, for a given task to be performed, with respect to individual assistance needs and comfort metrics.
-
Generat ing optimal assistance from human-robot simulation
: Candidate mechatronic solutions (from second axis) can be evaluated on simulations, to find best trajectories of assistance
, because general metrics such as the manipulability or a predefined workspace do not sufficiently capture the robot behavior.
The architecture (from second axis) and control optimization is the core principle of co-design [Bravo22, Fadini22]. Based on a coupled simulation of the human and exoskeleton, we will generate optimal assistance that leverage the models of human biomechanical variability (from the first axis)
. The optimal control problem will be formulated that the assistance adapts to the variation in motion-force capacity of the user, as well as is robust to variability in human morphology. We w ill use efficient non-linear MPC to solve this optimal control problem [Jordana25] while predicting future human-robot behavior from the models. Sensitivity approaches to MPC will be used to study the impact of variability or uncertainties in the human model parameters on the assistance
[Belvedere25].

Skills

The candidate should have solid skills in kinematic and dynamic modeling, data analysis, programming (C++ or Python), robotic control
, and mechatronic design.
Additional experience in biomechanics, statistics, exoskeleton design, or optimal control would be appreciated.

This listing was collected from a public source and is reproduced here for information only. Always confirm the details on the original posting before applying.
View the original posting

Similar positions