Dependency

• ROS
• V-REP
• Other tools: ros-kinetic-hector-slam, opencv-python, dlib

Code Structure

cd ./robot_project
tree -L 2
.
├── launch
│   └── final_project.launch   # all in one launch file
├── rviz_cfg                   # rviz config file for demo,
│   └── demo.rviz              # auto loaded by final_project.launch
├── src
│   ├── image_localization     # node image_localization
│   ├── laser_mapping          # node laser_mappting
│   └── visual_servoing        # node visual_servoing
└── ttt
    └── env_modified.ttt       # env file used by V-REP

Run

# terminal 1 run roscore
roscore
# terminal 2 start vrep simulation env
cd $VREP_dir 
./vrep.sh
# terminal 3 run all with one laucn file
cd ./robot_project
catkin_make
roslaunch launch/final_project.launch

The interface is shown as bellow:

Nodes

There are three ROS nodes in this project: laser_mapping, image_localization and visual_servoing. And these nodes are responsible for respective functions as follows:

1. in node laser_mapping, we use arrow keys to move the robot and simultaneously build the 2D map through hector slam provided with laser scan data,
2. in node image_localization, the robot detects and localize the images on the wall and mark them in the rviz, and also this node checks which area the robot enters in,
3. in node visual_servoing, it detects the yellow moving ball and if it occurs, it will switch from key controlling mode to auto tracking mode, and simultaneously the laser scan sensor turns off.

xxxxxx