This project is a Mapping and Deep Learning Object Recognition Project for use with the Turtlebot. (June - 8th September 2017)

• Produce a 2D occupancy grid map and 3D point cloud using RTAB-Map
• Autonomously navigate an unknown environment
• Detect and identify objects in a room
• Calibrate camera to mask textures

• Turtlebot 2

• A USB cable (that works, ensure it does) to connect the Kobuki Base and Laptop

• A Laptop running Ubuntu 16.04 and ROS Kinetic

• A camera:

  • Xbox Kinect v2 connected to the 12v 5 amps socket on the Turtlebot. (Please note the existing cable for this is poor and will need a permanent solution with proper parts.)

  • Zed Camera connected to the laptop

0. Install Turtlebot packages (replace for kinetic, some will not work) following instructions found here:

   http://wiki.ros.org/turtlebot/Tutorials/indigo/Turtlebot%20Installation

1. Install RTABMAP-ros following instructions found here:

   https://github.com/introlab/rtabmap_ros

2. Install the Exploration package:

cd ~/catkin_ws/src
git clone https://github.com/bnurbekov/Turtlebot_Navigation
cd ..
catkin_make

3. Install the Google Cloud SDK following instructions found here:

   https://cloud.google.com/sdk/downloads

4. Enable the SDK for the Google Vision API and install the client library following instructions found here:

   https://cloud.google.com/vision/docs/reference/libraries

5. Clone and catkin_make in a catkin workspace:

cd ~/catkin_ws/src
git clone https://github.com/mcgeorgiev/terrapin-ros
cd ..
catkin_make

6. (OPTIONAL) Tensorflow will need to be trained and placed in ~/terrapin-ros/src/tensorflow/tf_files following instructions found here:

   https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#0

a) Kinect v2:

1. https://github.com/OpenKinect/libfreenect2
2. https://github.com/code-iai/iai_kinect2

b) Zed Camera:

1. https://www.stereolabs.com/blog/index.php/2015/09/07/use-your-zed-camera-with-ros/ (Including the SDK instruction)

cd ~/catkin_ws/src/terrapin-ros 
pip install –r requirements.txt

0. Ensure that the catkin workspace directory is sourced for all terminals used. Usually:

source ~/catkin_ws/devel/setup.bash

1. Run the launch file specific to your camera, either:

roslaunch terrapin-ros kinect.launch
OR
roslaunch terrapin-ros zed.launch

This will launch the turtlebot_bringup, rtabmap_ros, rtabmap visualisation, specific camera node and depthimage_to_laserscan nodes.

2. Run the object detection programme:

roslaunch terrapin-ros stream.py kinect2
OR
roslaunch terrapin-ros stream.py zed

3. Run RViz:

...

4. Run the frontier exploration nodes:

rosrun final_project control.py
rosrun final_project mapping.py

• Turtlebot should start mapping! However autonomous navigation can be replaced with tele-operation. Replace step 4) with: roslaunch turtlebot-telep keyboard.launch

• A calibration tool can be ran which will create a text file with calibration details. This will mask out the selected area. (E.g. flooring) Run: python calibration.py and point at an area and press ‘q’ or ‘p’.