Skip to content

Commit

Permalink
Merge pull request #1 from dancodery/patch-1
Browse files Browse the repository at this point in the history
Patch 1
  • Loading branch information
dancodery authored Jan 17, 2019
2 parents feb8e82 + cd74a0b commit 16dae4f
Show file tree
Hide file tree
Showing 2 changed files with 4 additions and 4 deletions.
2 changes: 1 addition & 1 deletion notebooks/Self-Driving Cars [Part 1- The ALV].ipynb
Original file line number Diff line number Diff line change
Expand Up @@ -610,7 +610,7 @@
}
],
"source": [
"interact(apply_alv_vision, t = (40, 130), thresh = (-64, 64))"
"interact(apply_alv_vision, t = (0, 10), thresh = (-64, 64))"
]
},
{
Expand Down
6 changes: 3 additions & 3 deletions notebooks/Self-Driving Cars [Part 2 - ALVINN].ipynb
Original file line number Diff line number Diff line change
Expand Up @@ -672,7 +672,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"- Our network has learned to intrepret driving images based ONLY on steering angle! \n",
"- Our network has learned to interpret driving images based ONLY on steering angle! \n",
"- For this data (track data with only left turns), our network has learned a set of lane marker detectors!\n"
]
},
Expand All @@ -688,7 +688,7 @@
"metadata": {},
"source": [
"- For a number of years from the late 1980s to the early 1990s, ALVINN was the most successfull vision algorithm for autonomous driving at CMU, and possibly on the planet.\n",
"- Faster computers enabled faster driving, and more sophisiticed training and data augmentation procedures allowed ALVINN to learn quickly and effectively across many environments."
"- Faster computers enabled faster driving, and more sophisticated training and data augmentation procedures allowed ALVINN to learn quickly and effectively across many environments."
]
},
{
Expand All @@ -703,7 +703,7 @@
"metadata": {},
"source": [
"- And quite recently, Nvidia implemented Pomerleau's idea using **Deep Neural Networks**\n",
"- Great writeup in thier publication [End to end learning for self-driving cars](https://images.nvidia.com/content/tegra/automotive/images/2016/solutions/pdf/end-to-end-dl-using-px.pdf)\n",
"- Great writeup in their publication [End to end learning for self-driving cars](https://images.nvidia.com/content/tegra/automotive/images/2016/solutions/pdf/end-to-end-dl-using-px.pdf)\n",
"- Pomerleau orinally trained on ~5 minutes of driving, Nvidia trained on 3000 hours\n",
" - That's 36,000 times the amount of data - **5 orders of magnitude** increase in dataset size in 30 years. \n"
]
Expand Down

0 comments on commit 16dae4f

Please sign in to comment.