“Give it a shot yourself - take this paper, install TensorFlow, Keras, and Python”
End to End Learning for Self-Driving Cars
NVIDIA 用 CNN 直接把摄像头像素映射为方向盘转角
论文训练一个卷积神经网络,把前置摄像头的原始像素直接映射为转向指令,仅以人类驾驶的方向盘转角作为训练信号。系统在有无车道线的本地道路、高速公路,以及停车场和未铺装路面等视觉指引不明确的区域都能行驶,内部会自动学到检测道路特征等处理步骤,在 NVIDIA DRIVE PX 上以 30 帧每秒运行。作者主张这种端到端方式最终会带来更好的性能和更小的系统。
HN 上的人在讨论自动驾驶时反复提它,作为只用摄像头图像、靠神经网络回归转向角的代表例子,也有人让别人照论文自己动手复现。
HN 书评
推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及“NVidia published a very interesting paper on this technique called "End-To-End Learning For Self-Driving Cars": End to End Learning for Self-Driving Cars. It's an enjoyable read.”
“NVidia, on the other hand, has been experimenting with using neural networks (deep learning CNNs specifically) to drive vehicles using only camera information:”
“But I think first we'll see cars utilizing tech as described in this paper:”
“This is following the number suggested in NVIDIA's paper[1], which cautions against overfitting.”
“NVIDIA has a paper on it[1]”
看全部 7 条 →
“In complex environment a work from Nvidia is notable”
- End to End Learning for Self-Driving Cars ↗dharbin · 2016-04-26 · 1 分
- ConvNet-Based “End to End Learning for Self-Driving Cars” ↗mashgin · 2016-04-27 · 13 分
- End to End Learning for Self-Driving Cars [pdf] ↗AndreasM · 2016-04-27 · 1 分