GTAV_SDC

Automatic Pedestrian Avoiding Vehicles in GTAV. It works ... when it does.

Presentation Video: https://www.youtube.com/watch?v=K5tWRnS-N8U&feature=youtu.be

Team Members

• Yilun Chen
  • Project Architect. Designed and implemented the Perception Engine.
  • Helped implementing and debugging the Keyboard Emulator and Data Visualization Engine.
• Jingxuan Wang
  • Designed and implemented the Keyboard Emulator and the Screen Capturer.
  • Designed customized byte packet format to efficiently receive command.
• Yida Wang
  • Designed and implemented the Data Visualization Engine.
  • Boosted capabilities of the engine to support live rendering of multiple objects.
• Bruce Qin
  • Designed and implemented the Data Ingestion Engine.
  • Significantly improved performance of Ingestion Engine by 8-10 times.

Contents from ECE4122/6122 Used In This Project

• Classes: Used by Data Visualization Engine to draw different objects.
• Sockets: Used by Data Ingestion Engine, Keyboard Emulator and Data Visualization Engine.
• Threading: Used by Data Visualization Engine.
• OpenGL: Used by Data Visualization Engine.

Installation

• GTAV_SDC runs on Windows only.

• Make sure to use python 3.7

• python -m pip install -r requirements.txt

• Install protobuf https://github.com/protocolbuffers/protobuf

• Download contents of https://github.com/tensorflow/models into tensorflow directory and add that directory to environment variables.

• (Optional) Install CUDA (for GPU acceleration)

• Install GTAV and run it in windowed mode, resolution 800*600, and pull it to the top left of the screen.

• Run .\GDIPlus_ScreenCap.exe to start taking screenshots.

• Run python3 .\obj_detection.py to start the Data Ingestion Engine and the Perception Engine.

• Run .\KeyBoardEmulation.exe to start the Keyboard Emulator.

• Run .\visualization_engine\release\visualization_engine.exe to start the Visualization Engine.

• GTAV_SDC should be fully up and running. Try running into pedestrians and watch GTAV_SDC save them all!