smoke

ROS package for smoke detection and recognition with Turtlebot.

This package is not sufficient for performing the task.

Prerequisites

• ROS kinetic
• darknet
• darknet_ros
• OpenCV (>=3.3.0)
• rbx1 (optional)
• Qt5 (optional)

Usage

1. Build libraries and executables (within an ROS workspace)

   catkin_make -DCATKIN_WHITELIST_PACKAGES="smoke"
   

   If you have Qt5 installed, a gui would be built for simple visualization (as follows).

   The build may not finish once. Please try again.

2. Run

   • launch webcam using usb_cam.
   • launch darknet_ros

     roslaunch darknet_ros darknet_ros.launch
     

   • launch nodes of this package
     • launch darknetDetector node

       rosrun smoke darknet_svm_node

     • launch svm server node

       rosrun smoke svm_server

     • launch kobuki alarm node

       rosrun smoke kobuki_actSrv.py

     • launch alarm node

       rosrun smoke alarm_sub.py

     • visualization

       rosrun smoke appviz_node

launch files

• smoke.1.launch

  test with real smoke and real robot

• darknet_ros_smoke.launch

  launch darknet_ros

• test_darknet.launch

  test the object detection framework

• test_act.launch

  test the action server for robot pose inference

• test_srv.launch

  test the service '/kinectdev/smoke/smoke_svm'

You can modify the .launch file to specify which nodes to run. Also modify the config file.