Development of a model reference computed torque controller for a human lower extremity exoskeleton robot

Abstract

Exoskeleton robot–based neurorehabilitation has received a lot of attention recently due to positive evidence supporting its ability to provide different forms of physical therapy and in helping evaluate the patient recovery rate accurately. The performance of exoskeleton robot–based physical therapy depends on the accuracy of the motion control system. While the computed torque control scheme based on inverse dynamics is ideal from a theoretical perspective, the stability and tracking performance strongly depends on the model accuracy. Expecting a deterministic payload for a rehabilitation robot is impractical, which makes the computed torque controller unrealistic for such an application. In this article, a 7-degree-of-freedom human lower extremity dynamic model is developed using the Lagrange method and a novel Model Reference Computed Torque Controller is utilized for control. The computed torque controller is used to estimate the joint torque requirements for tracking a trajectory. Calculated joint torques are applied to a similarly structured plant with different parameters. The deviation of the plant from the model is calculated. A proportional–integral–derivative controller is employed to force the plant to behave like the robot model. A realistic friction model is incorporated to simulate joint friction in the plant. The stability and tracking performance of the control system is presented for sequential as well as simultaneous joint movements. To verify the robustness of the developed controller, analysis of variance statistical technique is used.

Constrained control methods for lower extremity rehabilitation exoskeleton robot considering unknown perturbations

“…Remark 1: For a given constraint, 𝐴 and 𝑏 are determined. There is at least one solution for 𝑥̈ in equation (6), then equation ( 6) can be seen as continuous [30].…”

“…Theorem 1 (U-K theory): Assumptions 1-2 are satisfied, an unconstrained mechanical system is subjected to constrains as shown in equation (6), then the constrain force 𝑄 𝑐 is shown as [10] 𝑄 𝑐 = 𝑀 1 2 ⁄ (𝐴𝑀 −1 2 ⁄ ) + × (𝑏 + 𝐴𝑀 −1 (𝐶 + 𝐺)). (7)…”

“…Clinical medical researches, meanwhile, have shown that human nervous system can reestablish synaptic connections through rehabilitation training [3]. In recent years, researchers have attached remarkable attention to LERER and conducted numerous studies in related fields [4][5][6].…”

Radial basis function‐based exoskeleton robot controller development

“…A linearised plant can be controlled by any type of linear control algorithm. The following are some examples of modelbased robot schemes that are often utilised in robot control applications: computer torque control [15], adaptive control [16], model reference computed torque control [17] and sliding mode control [18]. As it is impossible to precisely quantify the payload of an exoskeleton robot, the robot controller needs to be strong enough to handle parametric variations.…”

“…A robot can be controlled by linear, non-linear, robust, optimal and adaptive control systems. Most established control approaches [15][16][17][19][20][21][22] are based on model-based control strategies. For linearisation and control, different parts of the robot's dynamic equation of motion are used.…”