Boubaker Daachi

A robust adaptive control of a parallel robot

The work presented in this article deals with the robust adaptive control tracking of a 6 degree of freedom parallel robot, called C5 parallel robot. The proposed approach is based on the coupling of sliding modes and multi-layers perceptron neural networks (MLP-NNs). It does not require the inverse dynamic model for deriving the control law. The MLP-NN is added in the control scheme to estimate the gravitational and frictional forces along with the non-modelled dynamic effects. The nonlinearity problem, present in neural networks, is resolved using Taylor series expansion. The proposed approach allows to adjust the parameters of neural network and sliding mode control terms by taking into account a reference model and the closed-loop stability in the Lyapunov sense. We implemented our approach on the C5 parallel robot of LISSI laboratory and performed experiments to observe its effectiveness and the robust behaviour of the controller against external disturbances.

Adaptive control of a human-driven knee joint orthosis

The paper concerns the control of a lower limb orthosis acting on the knee joint level. Therefore, a model of the shank-orthosis system is given considering the human effort as an external torque acting on the system. A model reference adaptive control law is developed and applied to the orthosis in order to make the system (shank-orthosis) track a desired trajectory predefined by a rehabilitation doctor. The main advantage of this control law is the on-line parameters regulation allowing to ensure the best performance of the system. A Lyapunov-based analysis is performed to prove the input-to-state stability of the orthosis with respect to a bounded human torque. The performance of the system is then shown through some simulations.

Adaptive controller based on uncertainty parametric estimation using backstepping and sliding mode techniques: Application to an active orthosis

In this paper we propose an adaptive controller based on sliding mode and backstepping approaches. The system to be controlled is an exoskeleton used for the rehabilitation of the knee joint. The person wearing the exoskeleton is healthy, sitting and performs movements of flexion/extension. This kind of movement is usually applied by the doctor. The parameters of the dynamic model of the overall system (knee & exoskeleton) are considered unknown. The proposed controller is stable and robust against bounded external disturbances. The experimental results show the effectiveness of the proposed approach. The tests have also been validated when the wearer applies a resistive/assistive torque.

A robust adaptive neural controller to drive a knee joint actuated orthosis

This paper presents a robust adaptive control of an actuated orthosis intended to assist the lower limb movements of dependent persons. The proposed controller, based on a MultiLayer Perception Neural Network (MLPNN) and considered as a black-box, does not require the dynamic model of lower limbs/orthosis. A neural identification is used to extract the principal components of the MLPNN input vector. The MLPNN is used to compensate the dynamic effects arising from the interaction between the human lower limb and the orthosis. MLPNN weights are adjusted online according to an adaption algorithm based on the Lyapunov analysis. Experiments, carried out on a healthy subject, show the good performance of the proposed controller in terms of trajectory tracking and robustness against external disturbances.

MLPNN adaptive controller based on a reference model to drive an actuated lower limb orthosis

In this paper we propose to drive an actuated orthosis using an adaptive controller based on a reference model. It is not necessary to know all the functions of the dynamic model. Needing only the global structure of the dynamic model, we use a specific adaptive controller to obtain good performance in terms of trajectory tracking both in position and in velocity. A Multi-Layer Perceptron Neural Network (MLPNN) is used to estimate dynamics related to inertia, gravitational and frictional forces along with other unmodeled dynamics. The Lyapunov formalism is used for stability study of the system (shank+orthosis) in closed loop and to determine adaptation laws of the neural parameters. To treat the non-linearties related to the MLPNN, we have used first order Taylor series expansion. Experimental results have been obtained using a real orthosis worn by an appropriate dummy. Several tests have been realized to verify the effectiveness and the robustness of the proposed controller. For instance, our proposed orthosis model has given robust tracking performance under assistive as well as resistive forces.

Adaptive observer based on MLPNN and sliding mode for wearable robots

This paper deals with the design of an adaptive observer based both on a Multi-Layer Perceptron Neural Network (MLPNN) and a sliding mode technique. Its main objective is to construct the complete state of a given exoskeleton worn by a human subject. The observer we propose in this paper can be used for any application: rehabilitation, assistance, etc. The dynamic model of the global system composed of the exoskeleton and the human is complex and supposed completely unknown. The MLPNN chosen for its characteristic of universal approximation has been used here to identify the unknown dynamic. Its parameters have been adjusted by taking into account the structure of the dynamic model of the considered system and the closed-loop stability based on Lyapunovs approach. A Taylor series expansion allows resolving the non-linearity problem present in the MLPNN. Besides the fact that the proposed adaptive observer can be integrated in a control scheme, it also allows us to study the behavior of the exoskeleton before any application on the human subject. The proposed study has been validated both in simulation and in experimentation. The obtained results show the effectiveness of the proposed adaptive approach.

RBF-Based Neuro-Adaptive Controller for a Knee Joint Rehabilitation Orthosis

In this paper, we address a knee joint orthosis control for rehabilitation purposes. Only the structure of the system’s dynamic model is supposed to be known. Inertia of the knee-shank-orthosis system is identified on-line using an adaptive term. In order to approximate all of the other dynamics (viscous and solid frictions, gravity related torque, etc.), we use an RBF Neural Network (RBFNN) with no off-line prior training. Adaptation laws of the neural parameters and the inertia adaptive term are derived from the closed loop system’s overall stability study using Lyapunov’s theory. The study considers three cases: wearer being completely inactive or applying either a resistive or an assistive torque. Simulation results and conducted analysis show the effectiveness of the proposed approach.

Impact-based contextual service selection in a ubiquitous robotic environment

Context has a crucial importance in the way actions are perceived and done, especially in ubiquitous robotics where context is rich and subject to substantial variations. Given that service selection focuses on the nonfunctional performance of services, it must be tightly related to the context. Unfortunately, as far as we know, previous works have not effectively considered this relation. First, most of the existing selection models rely on Quality of Service (QoS) parameters that have been estimated according to the previous executions. However, two consecutive executions might occur in two very different contexts and then behave differently. Thus, this paper argues that these QoS parameters should be predicted from context. Finally, the aggregation of these QoS parameters into a score reflects the expectations on a service; it should also be context-dependent. In this article, a solution addressing these points is proposed for auxiliary services. Auxiliary services assist another service during its execution, usually by delivering a data stream. Instead of focusing on their individual performances, selection considers their impact on the assisted service. We propose to obtain this model through a multilayer perceptron under batch learning. Thus, focus is given to the sample generation. This model is validated in a ubiquitous robotic scenario involving a localization service selection.