Efficient learning of constraints and generic null space policies
Efficient learning of constraints and generic null space policies
复制标题
有效学习约束和通用零空间策略
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
S. Vijayakumar
中科院分区：
文献类型：
--
作者：
L. Armesto;Jorren Bosga;V. Ivan;S. Vijayakumar
一个大类的运动可以分解成一个运动任务和零空间的政策受到一组约束。当从演示中学习这种运动时，我们的目标是在不同的看不见的约束条件下实现泛化，并在保持低计算成本的同时提高对噪声的鲁棒性。存在用于学习运动策略和约束的各种方法。这些技术的有效性已被证明在低维场景和简单的运动。在本文中，我们提出了一种快速，准确的方法来学习约束的观察。这种新的提法的问题允许的约束学习方法与政策学习方法相结合，以提高政策学习的准确性，这使我们能够学习更复杂的运动。我们展示了我们的方法，通过学习一个复杂的表面擦拭政策，在7自由度的机器人手臂。
A large class of motions can be decomposed into a movement task and null-space policy subject to a set of constraints. When learning such motions from demonstrations, we aim to achieve generalisation across different unseen constraints and to increase the robustness to noise while keeping the computational cost low. There exists a variety of methods for learning the movement policy and the constraints. The effectiveness of these techniques has been demonstrated in low-dimensional scenarios and simple motions. In this paper, we present a fast and accurate approach to learning constraints from observations. This novel formulation of the problem allows the constraint learning method to be coupled with the policy learning method to improve policy learning accuracy, which enables us to learn more complex motions. We demonstrate our approach by learning a complex surface wiping policy in a 7-DOF robotic arm.