Configuration Estimation for Accurate Position Control of Large-Scale Soft Robots
Configuration Estimation for Accurate Position Control of Large-Scale Soft Robots
复制标题
DOI：
10.1109/tmech.2018.2878228
复制
发表时间：
2019-02-01
影响因子：
6.4
通讯作者：
Killpack, Marc D.
中科院分区：
文献类型：
--
作者：
Hyatt, Phillip;Kraus, Dustan;Killpack, Marc D.
在机器人技术中，有一个重要的趋势是探索被动顺应和低惯性系统，可以安全地与环境接触。本文定义了许多与开发软，气动驱动，充气机器人的有效控制相关的问题，并提出了一种方法来解决这些问题的一个子集。我们表明，我们可以获得一个全球性的测量方向为一个给定的软机器人链接使用两种不同类型的传感器（不包括运动捕捉）。给定取向测量，可以估计软致动器和关节的相对配置。为了验证控制末端执行器位置和方向的能力，我们提出了一种校准两个不相关测量系统坐标系的新方法。然后，使用我们的配置估计方法之一，我们证明了它的可行性，通过执行简单的行为与大型（约1.5米长）的软机器人平台连接到K-Rex漫游者在美国宇航局艾姆斯在户外环境中。我们的研究结果还显示了软机器人运动学校准的重要性和软机器人对结构中简单扰动的敏感性，如放气和再充气。最后，我们表明，末端执行器的错误可以显着减少做一种形式的伺服。总之，我们的方法和演示表明，有效的软机器人配置估计和控制的大型软机器人能够执行操作任务。
There is a significant trend in robotics of exploring passively compliant and low inertia systems that can safely make contact with the environment. This paper defines many of the problems associated with developing effective control of soft, pneumatically actuated, inflatable robots, and proposes an approach to solving a subset of these problems. We show that we can obtain a global measurement of orientation for a given soft-robot link using two different types of sensors (not including motion capture). Given the orientation measurement, it is possible to estimate relative configurations of the soft actuators and joints. In order to validate the ability to control position and orientation at the end effector, we show a new method for calibrating the coordinate frames of two unrelated measurement systems. Then, using one of our configuration estimation methods, we demonstrate its viability by performing simple behaviors with a large-scale (approximately 1.5 m long) soft-robot platform attached to the K-Rex rover at NASA Ames in an outdoor environment. Our results also show the importance of soft-robot kinematic calibration and the sensitivity of a soft robot to simple perturbations in the structure like deflation and reinflation. Finally, we show that end effector error can be significantly reduced by doing a form of servoing. In summary, our approach and demonstrations show effective soft-robot configuration estimation and control for large-scale soft robots capable of performing manipulation tasks.