丝驱动连续体机器人的无模型自适应控制

Model-Free Adaptive Control for Tendon-Driven Continuum Robots

连续体机器人由于具有高灵活度的特点,被广泛用于微创外科手术、航空、核工业等领域.针对连续体机器人在狭窄、受限环境下的建模误差导致控制精度较低的问题,提出了一种基于无模型自适应控制(MFAC)的连续体机器人末端位置控制方法.首先,对连续体的运动学模型和基于模型的控制方法(MBF)进行了研究,分析了建模误差对连续体机器人运动精度的影响;随后根据连续体机器人的驱动输入与位置输出,实时估计连续体机器人的伪雅可比矩阵,以减弱迟滞、模型误差、外部负载等干扰带来的影响;最后在连续体机器人平台上进行了轨迹跟踪实验,探究了自由状态下连续体机器人末端点的轨迹,随后改变轨迹参数使连续体机器人在不同半径的圆轨迹和不同顶角大小的三角轨迹下进行了轨迹跟踪实验和实验结果比较,并将本文所提出的方法与常见的基于模型的控制方法进行了比较,证明了本文方法具有较好的精度与抗干扰能力.实验结果表明:连续体机器人沿着半径为40 mm的圆形轨迹及其内接三角轨迹的均方根误差的平均值分别为2.31 mm和2.448 mm,具有良好的控制精度,且在运动过程中均方根误差值保持稳定;本文方法的轨迹均方根误差随着连续体运动范围增大而增大,运动精度保持在可接受的范围内;本文方法与基于模型的控制方法相比,可很好地抵抗外部干扰,提高连续体机器人的控制精度.

Continuum robots have been increasingly employed in various fields,such as minimally invasive surgery,aerospace,and nuclear industry,due to their high flexibility.A model-free adaptive control-based position control method for continuum robots is proposed in this paper to improve their tracking accuracy in constrained environments.First,the kinematic model-based control method of continuum robots,and effects of the modeling error on the distal positioning accuracy were studied and analyzed.Next,the pseudo-Jacobian matrix of the continuum robot was estimated in real time according to its control input and distal position output signals to reduce external disturbances,such as hysteresis,modeling error,and other interferences.Finally,experiments on the developed continuum robot platform were conducted,and the tracking error for two different trajectories was analyzed in the free space.The comparison between the proposed method and the model-based feedback method indicated that this method had better precision and anti-interference ability.The results showed that the root mean squared error(RMSE)of the continuum robot along the circular and triangular trajectories were 2.31 mm and 2.448mm,respectively,indicating the excellent control accuracy of the proposed method and a stable root mean square error value during the tracking movement.The RMSE of the proposed method increases with the expansion of the motion range and remains stable within an acceptable error range. The proposed method considerably resisted external disturbances and improved the accuracy of the control of continuum robots.