穿过房间有多难？--训练（循环）神经网络以操纵UAV
How hard is it to cross the room? -- Training (Recurrent) Neural Networks to steer a UAV
摘要：
这项工作探讨了在高级别导航任务的背景下，基于前视摄像头的输入，使用（循环）神经网络操纵无人机的可行性。我们建立了一个通用的框架，用于训练网络执行基于模仿学习的导航任务。它可以应用于空中和陆地车辆。作为概念验证，我们将其应用于模拟环境中的无人机（UAV），学习穿越包含许多障碍物的房间。到目前为止，只有前馈神经网络（FNN）已被用于训练无人机控制。为了应对更复杂的任务，我们提出使用递归神经网络（RNN），并成功地训练了一个LSTM（长短期记忆）网络来控制无人机。基于视觉的控制是一个序列预测问题，以其高度相关的输入数据而闻名。这种相关性使得训练网络变得困难，尤其是RNN。为了克服这个问题，我们在训练过程中研究了另一种采样方法，即窗口截断反向传播时间（WW-TBPTT）。此外，端到端培训需要大量数据，而这些数据通常是不可用的。因此，我们比较了仅重新训练全连接（FC）和LSTM控制层与端到端训练的网络的性能。执行相对简单的穿越房间的任务已经揭示了训练神经控制网络的重要指导方针和良好实践。不同的可视化有助于解释学到的行为。
英文摘要：
This work explores the feasibility of steering a drone with a (recurrent) neural network, based on input from a forward looking camera, in the context of a high-level navigation task. We set up a generic framework for training a network to perform navigation tasks based on imitation learning. It can be applied to both aerial and land vehicles. As a proof of concept we apply it to a UAV (Unmanned Aerial Vehicle) in a simulated environment, learning to cross a room containing a number of obstacles. So far only feedforward neural networks (FNNs) have been used to train UAV control. To cope with more complex tasks, we propose the use of recurrent neural networks (RNN) instead and successfully train an LSTM (Long-Short Term Memory) network for controlling UAVs. Vision based control is a sequential prediction problem, known for its highly correlated input data. The correlation makes training a network hard, especially an RNN. To overcome this issue, we investigate an alternative sampling method during training, namely window-wise truncated backpropagation through time (WW-TBPTT). Further, end-to-end training requires a lot of data which often is not available. Therefore, we compare the performance of retraining only the Fully Connected (FC) and LSTM control layers with networks which are trained end-to-end. Performing the relatively simple task of crossing a room already reveals important guidelines and good practices for training neural control networks. Different visualizations help to explain the behavior learned.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
