Shape-centric modeling for control of traveling wave rectilinear locomotion on snake-like robots
Shape-centric modeling for control of traveling wave rectilinear locomotion on snake-like robots
复制标题
DOI：
10.1016/j.robot.2019.103406
复制
发表时间：
2020-02-01
影响因子：
4.3
通讯作者：
Vela, Patricio A.
中科院分区：
文献类型：
--
作者：
Chang, Alexander H.;Vela, Patricio A.
行波直线步态的细长，连续的机构被建模为一个周期性变化的骨干曲线。步态形状被表示为相对于平均身体曲线和相关联的刚性附接的身体框架的平面偏差。人体-地面接触模式和其他几何性质的计算外强迫方便地定义相对于这个平均身体曲线。引入一个身体地面滚动摩擦模型允许的运动控制方程推导出封闭的形式。将恒定曲率施加到平均身体中实现了转弯运动，从而实现了转弯控制。系统动力学的重复数值积分有助于构造控制到动作的映射，表征相对于步态参数空间的稳定系统行为。控制到动作的映射将这个复杂的动力系统简化为一个运动学单一模型，其中反馈跟踪策略是很好理解的。为了说明其效用，它被应用在一个轨迹规划和跟踪框架周围的障碍物的运动。使用该框架，机器人蛇行使行波直线步态成功地计划可行的轨迹和穿越非平凡的障碍物安排，以达到指定的目标位置。(C)2019 Elsevier B. V.版权所有。
A traveling wave rectilinear gait for elongated, continuous bodies is modeled as a cyclically-varying backbone curve. The gait shapes are represented as planar deviations relative to an average body curve and an associated, rigidly-attached body frame. Body-ground contact patterns and other geometric properties integral to computation of external forcing are conveniently defined with respect to this average body curve. Introducing a body-ground rolling friction model permits the controlled equations of motion to be derived in closed form. Incorporating a constant curvature into the average body realizes turning movements, and hence turning control. Repeated numerical integration of the system dynamics facilitates construction of a control-to-action mapping, characterizing steady system behavior with respect to the gaits parameter space. The control-to-action map reduces this complex dynamical system to a kinematic unicycle model for which feedback tracking strategies are well understood. To illustrate its utility, it is applied in a trajectory planning and tracking framework for locomotion around obstacles. Using the framework, a robotic snake exercising the traveling wave rectilinear gait successfully plans feasible trajectories and traverses non-trivial obstacle arrangements to reach specified goal positions. (C) 2019 Elsevier B.V. All rights reserved.