gather data

Gathering Training Data

In order to keep the training data organized and in the same format so we call can benefit from your AWESOME driving skillz we've added some functions to the original Donkeycar. It is build in such a way that the commands like python manage.py drive ... are the same. Just with a tiny by of automated conveniences.

Tracking data meta info

The info.json file is used to keep track of various information about each training run. The count and date fields are filled automatically by the script, but the others need to be set by you before you begin gathering data. Feel free to add more information as you see fit.

{
  "count"       : -1,
  "date"        : "SET AUTOMTICALLY",
  "car name"    : "dingo1",
  "driver name" : "tall-josh",
  "tape"        : "thick green",
  "floor"       : "timber",
  "location"    : "Buckingham Palace",
  "centre line" : false,
  "lane width"  : "40cm"
}

Gathering data

This is exactly the same as Donkeycar Get Driving so I'm not going to just repeat it. Dingo changes occur as soon as you kill ctrl+c the driving script.

You will be prompted with:

What would you like to do with this data?
(z)ip and ship, (d)elet, (l)eave me alone! : 

Assuming you have the info.json file defined above.

1. zip and ship will (by default):

• make a directory at data/Buckingham_Palace_YYYY-MM-DD_hh:mm:ss
• fill the count and date fields in the info.json
• copy info.json to the new directory
• copy the first image, 0_cam-image_array_.jpgto the new directory
• create a .zip of the tub, call it tub.zip and put it in the new directory
• delete all the contents of except for meta.json
• Attempt to rsync the data back to your primary computer for training. (more info here)
• the final file structure should look like:

ls data/sample_tub
0_cam-image_array_.jpg  info.json   tub.zip

2. delete will:

• delete all the contents of except for meta.json

3. leave me alone:

• will do nothing, all data will remain untouched in tub directory

However, when you stop gathering data

Once you have enough data you're ready to Train