Model-Free Control Techniques for Robotic Exoskeletons

Model-free control (MFC) techniques circumvent the need for explicit dynamic models by employing real-time estimation of system behaviour. In the context of robotic exoskeletons, where complex kinematics, soft tissue interactions and varying user profiles pose significant modelling challenges, model-free approaches afford enhanced adaptability and robustness. Core MFC strategies for exoskeletons include ultra-local modelling, which approximates system dynamics via low-order differential relations; sliding mode control schemes that leverage switched control laws to tolerate uncertainties; intelligent proportional–integral control, incorporating online adjustment to counteract unmodelled dynamics; and adaptive neuro-fuzzy frameworks that approximate unknown functions through learning. These methods have been applied to both upper- and lower-limb exoskeletons for rehabilitation, assistance and strength augmentation, demonstrating rapid response, disturbance rejection and finite-time convergence in tracking desired motion trajectories. Practical implementations have spanned wearable exosuits for gait support, wheelchair-mounted arms for quadriplegic patients and powered orthoses for industrial augmentation. The global significance of this research lies in its potential to democratise advanced mobility aids and industrial augmenters by reducing sensor and calibration requirements, thereby accelerating clinical translation and field deployment.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Model-Free Control Techniques for Robotic Exoskeletons publication trend

The graph below shows the total number of articles in model-free control techniques for robotic exoskeletons across all publications each year (not limited to Nature Index journals).

Technical terms

Model-free control: A control approach that eschews explicit system models by estimating dynamics in real time via input–output data.

Ultra-local model: A low-order differential relation representing dominant system behaviour over a narrow time horizon.

Sliding mode control: A robust control technique that forces system states onto a predefined manifold by switching control laws.

Super-twisting algorithm: A higher-order sliding mode control law that reduces chattering while ensuring finite-time convergence.

Terminal sliding mode: A variant of sliding mode control achieving finite-time convergence to the sliding manifold via nonlinear surfaces.

Fixed-time convergence: A convergence property guaranteeing that system errors reach zero within a prescribed time bound independent of initial conditions.

Disturbance observer: An estimator that reconstructs external disturbances and unmodelled dynamics to compensate in the control loop.

Adaptive neural network: A learning-based approximation mechanism that adjusts weights online to estimate unknown system functions.

Fuzzy logic: A rule-based approach handling uncertainty by reasoning with linguistic variables and membership functions.

References

1. Adaptive Neural Network-Based Fixed-Time Tracking Controller for Disabilities Exoskeleton Wheelchair Robotic System. Mathematics (2022).
2. Fuzzy-Based Fixed-Time Nonsingular Tracker of Exoskeleton Robots for Disabilities Using Sliding Mode State Observer. Mathematics (2022).
3. An Improved Super-Twisting Sliding Mode for Flexible Upper-Limb Exoskeleton. Actuators (2023).