Deep Active Localization
Deep Active Localization
复制标题
DOI：
10.1109/lra.2019.2932575
复制
发表时间：
2019-10-01
影响因子：
5.2
通讯作者：
Paull, Liam
中科院分区：
文献类型：
--
作者：
Gottipati, Sai Krishna;Seo, Keehong;Paull, Liam
主动定位包括生成机器人动作，允许它在参考地图内最大限度地消除其姿势1%的歧义。传统的方法使用信息理论标准来选择动作和手工制作的感知模型。在这项工作中，我们提出了一个端到端的微分学习方法，采取信息的行动，是完全在模拟训练，然后转移到真实的机器人硬件零细化。该系统由两个学习模块组成：用于感知的卷积神经网络和深度强化学习规划模块。我们在感知模型中利用了多尺度方法，因为使用强化学习采取行动所需的准确性远远低于机器人控制所需的准确性。我们证明了所得到的系统优于传统的方法，无论是感知或规划。我们还证明了我们的方法的鲁棒性，不同的地图配置和其他滋扰参数，通过使用域随机化训练。该代码已发布：https://github.com/montrealrobotics/dal，并与OpenAI健身房框架以及Gazebo模拟器兼容。
Active localization consists of generating robot actions that allow it to maximally disambiguate its pose 1% within a reference map. Traditional approaches use an information-theoretic criterion for action selection and hand-crafted perceptual models. In this work we propose an end-to-end differentiable method for learning to take informative actions that is trainable entirely in simulation and then transferable to real robot hardware with zero refinement. The system is composed of two learned modules: a convolutional neural network for perception, and a deep reinforcement learned planning module. We leverage a multi-scale approach in the perceptual model since the accuracy needed to take actions using reinforcement learning is much less than the accuracy needed for robot control. We demonstrate that the resulting system outperforms traditional approach for either perception or planning. We also demonstrate our approach's robustness to different map configurations and other nuisance parameters through the use of domain randomization in training. The code has been released: https://github.com/montrealrobotics/dal and is compatible with the OpenAI gym framework, as well as the Gazebo simulator.