Skip to content

Latest commit

 

History

History
31 lines (17 loc) · 1.13 KB

README.md

File metadata and controls

31 lines (17 loc) · 1.13 KB

VPilot

Scripts and tools to easily communicate with DeepGTAV. In the future a self-driving agent will be implemented.

Self-Driving Car

How it works

VPilot uses JSON over TCP sockets to start, configure and send/receive commands to/from DeepGTAV by using the Python DeepGTAV libraries.

dataset.py and drive.py serve as examples to collect a dataset using DeepGTAV and giving the control to an agent respectively.

How to use it

DATASET.PY : python dataset.py -d dataset/

Simulation configuration

weathers = ["CLEAR", "EXTRASUNNY", "CLOUDS", "OVERCAST", "RAIN", "CLEARING", "THUNDER", "SMOG", "FOGGY", "XMAS", "SNOWLIGHT", "BLIZZARD", "NEUTRAL", "SNOW" ]

hours = [0,4,8,12,16,20]

DATASET FORMAT:

dataset/weather/hour/timestamp_simulation/index_of_frames.pz

PLUGIN DEV

ADD _IS_VEHICLE_DAMAGED for instance in order to get vehicule damage.

See http://www.dev-c.com/nativedb/ for a full list of all the methods that can be used.