Spatial adaption of robot trajectories based on laplacian trajectory editing
Spatial adaption of robot trajectories based on laplacian trajectory editing
复制标题
基于拉普拉斯轨迹编辑的机器人轨迹空间自适应
DOI：
--
复制
发表时间：
2015
影响因子：
3.5
通讯作者：
Yoshihiko Nakamura
中科院分区：
文献类型：
--
作者：
Thomas Nierhoff;S. Hirche;Yoshihiko Nakamura
假设机器人轨迹是由一个高级规划或学习机制给出的，它需要适应动态环境的变化。在本文中，我们提出了一种新的方法来变形轨迹，同时保持其局部形状相似，这是基于离散拉普拉斯-贝尔特拉米算子。该方法可以很容易地扩展并涵盖多种变形技术，包括必须通过的固定路径点，避免碰撞的位置约束或多个机器人协调的合作操作方案。由于其较低的计算复杂度，它允许在局部和全局尺度上实时轨迹变形，并在线适应变化的环境约束。仿真说明了所提出的方法与其他已建立的轨迹相关方法（如人工势场或优先逆运动学）的直接结合。HRP-4类人机器人的实验成功地证明了它在复杂的日常生活任务中的适用性。
Assuming that a robot trajectory is given from a high-level planning or learning mechanism, it needs to be adapted to react to dynamic environment changes. In this article we propose a novel approach to deform trajectories while keeping their local shape similar, which is based on the discrete Laplace–Beltrami operator. The approach can be readily extended and covers multiple deformation techniques including fixed waypoints that must be passed, positional constraints for collision avoidance or a cooperative manipulation scheme for the coordination of multiple robots. Due to its low computational complexity it allows for real-time trajectory deformation both on local and global scale and online adaptation to changed environmental constraints. Simulations illustrate the straightforward combination of the proposed approach with other established trajectory-related methods like artificial potential fields or prioritized inverse kinematics. Experiments with the HRP-4 humanoid successfully demonstrate the applicability in complex daily-life tasks.