Quasi-Passive Human-Exoskeleton System Modelling

Christopher Caulcrick

The project, funded through an EPSRC Industrial Cooperative Award in Science & Technology (CASE) in partnership with McLaren Applied Technologies (MAT), involves virtually modelling human and exoskeleton together. Working closely with the modelling and simulation team at MAT, partially-passive human-exoskeleton systems are to be simulated for design and control optimisation.

Quasi-Passive Exoskeletons

Exoskeletons typically use multiple electric motors to actuate human joints at the hip, knee, and ankle - producing expensive, heavy, bulky, and power-hungry machines. Passively actuating knee and ankle joints, using mechanical elements such as linear/torsional springs/dampers/inerters and clutches, could significantly reduce cost, weight, size, and power consumption.

Configuring such elements and determining suitable mechanical characteristics is a challenging design stage which would traditionally involve building expensive prototypes. By instead modelling proposed designs in a virtual prototyping environment, time and money can be saved. This approach also allows designs to be modified and iterated much more easily.

Modelling

Taking inspiration from biomechanical models of human leg function and prior research in the field of quasi-passive prosthetics and exoskeletons, initial joint mechanism designs will be proposed. Mechanical linkages that constrain relative motion between lower limb regions will be considered. Designs can be parameterised in terms of mechanical characteristics, dimensions, and actuation requirements. Modelling will take place in SolidWorks, SimMechanics, and MapleSim.

A musculoskeletal model of the human body will be introduced to co-simulate behaviour of the interaction between human and exoskeleton. With the two models combined, a multi-objective optimisation of the design is possible with human-in-the-loop. Detailed characterisation of the design gait characteristics, joint dynamics/kinematics, ground reaction forces, and body interaction forces will take place.

Control of the system will also be simulated using information from sensor simulations and measurements. Machine learning techniques enable intelligent utilisation of data. Novel synergy of sensor modalities is proposed.

Outcomes

This project will provide increased understanding of exoskeleton design considerations and limitations. A product will be optimised novel design proposals for quasi-passive exoskeleton systems. A functional modelling, simulation, and optimisation method for future design proposals will be laid out, providing a useful tool for validating designs and characterising human-exoskeleton systems in detail, including controller and human-machine interface.