.png){kind=icon}
.png){kind=icon}
Return
Parameter Estimation and Q-Learning Based Adaptive Sensitivity Amplification Control for Cable-Driven Lower-Limb Exoskeleton
.png){kind=icon}
PRE
AI
DOI：10.1002/rnc.8005.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper presents a parameter estimation and an adaptive sensitivity amplification control (ASAC) for a cable driven lower-limb exoskeleton (CDLEX) to guarantee the tracking performance and enhance comfort in human-robot interaction (HRI). The cable-sheath actuator characterized by remote actuation capabilities and simplicity is applied to the lower-limb exoskeleton to establish a nonlinear integrated model. Aiming to identify the unknown parameters in this model, a novel adaptive parameter estimation framework driven by the extracted error information is proposed to improve the estimation veracity and convergence rate over to the traditional method. Moreover, a sensitivity amplification control (SAC) is adopted to maximize the sensitivity of the closed-loop system to external human-robot interaction (HRI) force/torque, where the stability and robustness are all analyzed. The proposed SAC dose not require the direct measurement of HRI in the SAC scheme. Therefore, it is possible to avoid installing the force/torque sensors between the human and the exoskeleton. To account for uncertainties in the interaction environment, such as the suddenly changing walking trajectories and individual gaits, the Q-learning algorithm is employed to realize online parameter tuning of SAC. Simulations and practical experiments are provided to illustrate the effectiveness of the proposed strategies.
Keywords:
human-robot interaction
lower-limb exoskeleton
parameter estimation
Q-learning
sensitivity amplification control
Journal
IF:
3.2
Papers:
6.9K
Citations:
1.4W