Adaptive robust control of soft bending actuators: an empirical nonlinear model-based approach

Abstract

Soft robotics, compared with their rigid counterparts, are able to adapt to uncharted environments, are superior in safe human-robot interactions, and have low cost, owing to the native compliance of the soft materials. However, customized complex structures, as well as the nonlinear and viscoelastic soft materials, pose a great challenge to accurate modeling and control of soft robotics, and impose restrictions on further applications. In this study, a unified modeling strategy is proposed to establish a complete dynamic model of the most widely used pneumatic soft bending actuator. First, a novel empirical nonlinear model with parametric and nonlinear uncertainties is identified to describe the nonlinear behaviors of pneumatic soft bending actuators. Second, an inner pressure dynamic model of a pneumatic soft bending actuator is established by introducing a modified valve flow rate model of the unbalanced pneumatic proportional valves. Third, an adaptive robust controller is designed using a backstepping method to handle and update the nonlinear and uncertain system. Finally, the experimental results of comparative trajectory tracking control indicate the validity of the proposed modeling and control method.

摘要

目 的

为软体机器人系统提供统一且完整的动力学建模方法, 并且基于建立的模型设计控制器, 以实现软体机器人的精确位置控制.

创新点

1. 提出了一种经验非线性模型及其辨识方法, 提高了气动软体机器人建模的精度; 2. 建立了不平衡气动比例阀的准静态流量模型, 实现了气动系统的动力学建模; 3. 基于模型, 设计了自适应鲁棒控制器, 实现了软体机器人的精确位置控制.

方 法

1. 将传统线性模型的参数设置为位置的函数, 使用泰勒展开、 系统滤波和最小二乘方法, 实现任意阶次的经验非线性模型辨识; 2. 对不平衡气动比例阀进行阀芯受力分析, 推导阀芯位置的准静态方程, 进而推导准静态流量模型; 3. 通过轨迹跟踪对比实验, 验证所提出的模型和控制器的有效性.

结 论

1. 实验结果表明, 仅使用滑模控制器, 就可以实现较高精度的轨迹跟踪, 这证明了所提建模方法的有效性; 2. 使用自适应鲁棒控制器, 并在传统滑模控制器的基础上在线更新参数, 可以有效提高轨迹跟踪精度.