minht57 / ROS_Basic_SLAM

BUILDING AN AUTOMATIC VEHICLE BASED ON STEREO CAMERA

BUILDING AN AUTOMATIC VEHICLE BASED ON STEREO CAMERA

Overview

This project focuses on using stereo camera to build a vehicle which can map the surrounding area and autonomosly navigate a to the pointed destination in the built map. The vehicle is a Turtlebot model and based on ROS (Robot Operating System). For a further purpose, the vehicle can be developed to go around a house doing diffrent tasks or explore unknown enviroment.

Algorithm

Dependencies

In order to run the project, the following dependencies are required:

• Ubuntu 16.04 here
• ROS Kinetic Kame here
• Catkin tool here
• ZED SDK (version 2.1) here
• Cuda Toolkit (version 8.0) here
• Firmware for Mobile base here
• ROS packages:
  • zed_ros_wrapper
  • depthimage_to_laserscan
  • laser_scan_matcher
  • gmapping
  • map_server
  • move_base
  • amcl
  • dwa_local_planner

Install ROS package dependencies:

sudo apt install ros-kinetic-<dependency_name>

Build and run the robot

Setup workspace:

git clone https://github.com/minht57/ROS_Basic_SLAM.git
cp -rf ROS_Basic_SLAM/src/* ~/catkin_ws/src/
sudo chmod 777 ~/catkin_ws/src/zed-ros-wapper/cfg/Zed.cfg
sudo chmod 777 ~/catkin_ws/src/depthimage_to_laserscan/cfg/Depth.cfg
cd ~/catkin_ws
catkin_make

Mapping:

Run the mapping:

roslaunch depthimage_to_laserscan gmapping.launch

Use RF controller to move the robot around. The map will be automatically built.

To see the map, launch Rviz and create visualization by adding 'Map' to the 'Display' console:

rviz

Save map:

rosrun map_server map_saver -f ~/catkin_ws/test_map

Raw map:

Edit the map with GIMP for better map:

Navigation:

Run navigation:

roslaunch navigation slam_amcl.launch

Send velocity commands to move base via serial port:

sudo chmod 777 /dev/ttyUSB0
roslaunch serial serial.launch

Place a box in the area to check object avoidance algorithm.

Open Rviz:

rviz

Result:

Youtube:

• Mapping: here
• Navigation: here