From 95ab08520c86c5ac8f0ba45c89d8675965013604 Mon Sep 17 00:00:00 2001 From: joe Date: Sun, 14 Jul 2019 22:21:01 -0400 Subject: [PATCH] Updated NN Network and fixed Readme Link --- README.md | 4 ++-- car/neural_network.py | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index 60d979b..db27911 100644 --- a/README.md +++ b/README.md @@ -7,7 +7,7 @@ PyTorch Python Neural Network Autonomous 1/10 Car for Nvidia Jetson Nano * **System Hardware** * Servo Driver https://www.amazon.com/dp/B014KTSMLA/ref=cm_sw_r_tw_dp_U_x_.PW4CbAD5YKP4 - * Jetson Nano https://store.nvidia.com/store?Action=DisplayPage&Locale=en_US&SiteID=nvidia&id=QuickBuyCartPage + * Jetson Nano https://developer.nvidia.com/embedded/jetson-nano-developer-kit * Camera https://leopardimaging.com/product/li-imx219-mipi-ff-nano/ * Wifi Chip https://www.amazon.com/gp/product/B01MZA1AB2/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1 * Wifi Attenna https://www.arrow.com/en/products/2042811100/molex @@ -29,4 +29,4 @@ PyTorch Python Neural Network Autonomous 1/10 Car for Nvidia Jetson Nano Install python requirements. Install ds4drv and connect PS4 Bluetooth controller to Ubuntu. Inside AutoCarJetsonNano/car, start main.py and drive car around. Images are captured when speed is > 0. After driving car around, offload images to remote computer for training and copy control_data.csv . Start autocar via Jupyter Lab or Notebook and train model via pytorch. Load model back to Nano after training and start main.py Press X to launch autopilot. -![Demo](media/JetsonNanoCarPytorch.gif) \ No newline at end of file +![Demo](media/JetsonNanoCarPytorch.gif) diff --git a/car/neural_network.py b/car/neural_network.py index 6bff376..493f533 100644 --- a/car/neural_network.py +++ b/car/neural_network.py @@ -18,7 +18,7 @@ def __init__(self): self.pool = nn.MaxPool2d(2, 2) self.fc1 = nn.Linear(64*3*3, 144) self.fc2 = nn.Linear(144, 1) - self.dropout = nn.Dropout(0.1) + self.dropout = nn.Dropout(0.15) def forward(self, x): # add sequence of convolutional and max pooling layers