Learning movement primitives for force interaction tasks
Learning movement primitives for force interaction tasks
复制标题
学习力交互任务的运动基元
DOI：
10.1109/icra.2015.7139639
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
Jochen J. Steil
中科院分区：
文献类型：
--
作者：
Jens Kober;M. Gienger;Jochen J. Steil
动觉教学是一种很有前途的以直观的方式获得机器人技能的方法。本文的重点是学习技能，不仅依赖于运动学，但也需要考虑相互作用的力量。我们提出了学习这种力相互作用技能的三个新概念。首先，我们使用接触信息从少量连续的动觉演示中确定片段。其次，我们将每个片段与运动原语联系起来，并确定其组成，即控制变量和参考框架，允许再现演示任务。最后，我们提出了一个概念来确定在繁殖过程中原语之间的转换。在一个拉盒和翻盒任务中对所提出的方法进行了评价，结果表明，该方法对不同几何形状的物体和不同物体排列情况具有很好的泛化能力。
Kinesthetic teaching is a promising approach to acquire robot skills in an intuitive way. This paper focuses on learning skills that do not solely rely on kinematics but also need to take into account interaction forces. We present three novel concepts towards learning such force interaction skills. Firstly, we determine segments from a small number of continuous kinesthetic demonstrations using contact information. Secondly, we associate each segment with a movement primitive, and determine its composition, i.e., the control variables and reference frames that allow to reproduce the demonstrated task. Lastly, we propose a concept to determine the transitions between the primitives during reproduction. The proposed methods are evaluated on a box pulling and flipping task, and show very good generalization abilities for objects with different geometries, and situations with different object arrangements.