Skip to content

Latest commit

 

History

History
37 lines (28 loc) · 1.32 KB

Readme.md

File metadata and controls

37 lines (28 loc) · 1.32 KB

#Drone Autonomous Searching System

A program that makes drone search in a closed area autonomously. By observing surroundings with a front-facing stereo camera (had no budget for 3 or 4 stereo cameras), it detects obstacles and avoids them in real-time. No foreknowledge of the environment is required. I use opencv cascade classifiers for detecting and searching for humans in a closed area.

The obstacle avoidance and human detection parts are written in c++, and then are wrapped as a nodejs binding for controlling drone using nodejs. (Since the third-party ARDrone2.0 nodejs client is much easier to program.)

##Requirememts

Hardware

Windows

  • Visual Studio 2013 (2015 is not supported by cuda 7.0)
  • Visual Studio 2012 redistributable x64
  • node v0.12.x
  • python 2.7 (for npm modules)
  • cmake-js v2.1.0
  • Nvidia Driver
  • CUDA 7.0
  • ZED SDK v0.9.2
  • OpenCV 2.4.9

###Add Path

  • Add (opencv_2.4.9 root)\build (absolute path) as system variable OPENCV_DIR
  • Add (opencv_2.4.9 root)\build\x64\vc12\bin (ausolute path) to Path

##Build & Run

 $ mkdir build && cd build
 $ cmake ..
 $ npm install		(to install dependent npm modules)
 $ npm run build		(to build)
 $ npm start			(to start)