Skip to content

Latest commit

 

History

History
29 lines (23 loc) · 1.55 KB

README.md

File metadata and controls

29 lines (23 loc) · 1.55 KB
Raw

Overview

The following sections provide info for some important files

Data collection

automatic_control.py and manual_control.py share the same code for lane extraction in lane_utils.py.

  • automatic_control.py is used for automatic data collection
  • manual_control.py is used for debugging and gathering map info where images are not saved

Data format

Lane labels are saved in the TuSimple format. lane_config.py specifies the row anchors used which can be changed to follow the CULane format.

Map info

map_info.py contains the following info for most available towns in CARLA

  1. the max number of lanes (including curbs) expected for each road id
  2. bad road ids where there are unfixable issues with the lane labels

This info is used to ensure the labels saved have no error. If a town is not in map info, it will not be used for data collection.

To add a new map to the collection,

  1. Generate the map using python manual_control.py -t {TownID} -m
  2. With the generated map as reference, use manual_control.py to drive around the town and add the required information in map_info.py using the same class structure as the other towns.

Dynamic weather

dynamic_weather.py is used to control how the weather changes in the simulation. The most important parameter is altitude, the equation that determines how the time of day changes over time here. Modify the sine function to change the range of time and ratio of day/night in the generated dataset.