التصميم الأمثل والتحكم في الهيكل الخارجي للطرف العلوي

Optimal Design and Control of an Upper-Limb Exoskeleton

Abstract

Smart upper limbs prosthetics play a significant role in enhancing mobility and independence for disabilities people with disabilities. These mechanical devices are utilized to provide functional movements of the human arm in various applications such as rehabilitation, sports, and technological innovation, one of these devices is a robotic arm exoskeleton. Therefore, it becomes essential to assess the dynamic response of robotic arm prosthetics. This study focuses on studying the dynamic response of a five-degree-of-freedom (5-DOF) prosthetic robotic arm.
The robotic arm system was modeled by using Lagrange-Euler method, resulting in five-second-order ordinary differential equations. To facilitate the optimization process required and obtain locally optimal system parameters. The nonlinear model was linearized around a specific operating point. A comparative analysis between the linear and nonlinear system responses was conducted to ensure that the linearized model can adequately represent and compensate for the real nonlinear system dynamics.
The arm parameters, including masses, damping coefficients, and stiffness values, were determined using a Lagrange multipliers optimization method with the objective of achieving a short settling time and minimal overshoot. Furthermore, eight-term exponential functions were employed to generate the desired displacement profiles for the hand, forearm, horizontal elbow, vertical elbow, and elbow angular motions while satisfying initial, intermediate, and final boundary conditions as well as elbow torque limitations.
The reference elbow torque was analytically computed based on the predefined displacement trajectories. To validate the proposed approach, numerical simulations were preformed using Monte Carlo simulation technique under a wide range of initial conditions and system parameter variations of (_-^+)6% from the nominal values. The obtained results demonstrated rapid convergence of all arm segments, achieving steady-state responses within 0.25 seconds with zero overshoot in both system displacements and elbow torque. Moreover, the steady-state errors remained extremely small, with a maximum elbow angular error of (4×10−3) degrees, a minimum vertical elbow displacement error of (−1×10−5) m in the vertical, and an elbow torque error as low as (-0.4) N.m across 1000 simulation trials.