Kinematics-Based Detection and Localization of Contacts Along Multisegment Continuum Robots
Kinematics-Based Detection and Localization of Contacts Along Multisegment Continuum Robots
复制标题
DOI：
10.1109/tro.2011.2175761
复制
发表时间：
2012-04-01
影响因子：
7.8
通讯作者：
Simaan, Nabil
中科院分区：
文献类型：
--
作者：
Bajo, Andrea;Simaan, Nabil
在本文中，我们提出了一种新颖的基于运动学的框架，用于沿多段连续体机器人的碰撞检测和接触位置估计。螺杆理论用于将螺杆运动偏差 (SMD) 定义为预期螺杆运动轴 (ISA) 与实际瞬时螺杆运动轴 (ISA) 之间的距离。预期 ISA 是根据机器人的无约束运动学模型计算的，而实际 ISA 是根据感觉信息计算的。通过监控 SMD 可以检测机器人沿线任意点与刚性环境的碰撞。通过最小化从连续体机器人的约束运动学模型获得的 ISA 和从传感器数据获得的 ISA 之间的 SMD 来估计接触位置。所提出的接触检测和定位方法仅需要每个连续体段相对于其自身基座的相对运动。该策略允许将这些算法直接推广到 n 段连续体机器人。该框架通过模拟和三段多主干连续体机器人上的实验进行评估。结果表明，碰撞检测算法能够检测任意段的单次碰撞、多个段发生的多次碰撞以及全臂约束。它还表明，可以在连续体机器人的任何位置估计接触位置，其精度优于段标称长度的 20%。我们相信这项研究将提高非结构化环境和密闭空间中的操作安全性。
In this paper, we present a novel kinematic-based framework for collision detection and estimation of contact location along multisegment continuum robots. Screw theory is used to define a screw motion deviation (SMD) as the distance between the expected and the actual instantaneous screw axis (ISA) of motion. The expected ISA is computed based on the unconstrained kinematics model of the robot, while the actual ISA is computed based on sensory information. Collisions with rigid environments at any point along the robot are detected by monitoring the SMD. Contact locations are estimated by the minimization of the SMD between the ISA that is obtained from a constrained kinematic model of the continuum robot and the one that is obtained from sensor data. The proposed contact detection and localization methods only require the relative motion of each continuum segment with respect to its own base. This strategy allows the straightforward generalization of these algorithms for an n-segment continuum robot. The framework is evaluated via simulations and experimentally on a three-segment multibackbone continuum robot. Results show that the collision-detection algorithm is capable of detecting a single collision at any segment, multiple collisions occurring at multiple segments, and total-arm constraint. It is also shown that the estimation of contact location is possible at any location along the continuum robot with an accuracy better than 20% of the segment nominal length. We believe this study will enhance manipulation safety in unstructured environments and confined spaces.