Config

Assuming desktop full verison of ros-kinetic has been installed, and a catkin workspace has been created at

/home/XXX/catkin_ws/

Follow the instruction at meta_ClosedLoopBench to clone all required catkin packages in the workspace:

https://github.com/ivalab/meta_ClosedLoopBench

Navigate to the dir of simulator package gazebo_turtlebot_simulaton. Adjust the catkin workspace in set_up_sim.sh:

export CATKIN_WS=/home/XXX/catkin_ws/

Then execute the auto setup script set_up_sim.sh (will be asked for sudo authorization):

./set_up_sim.sh

Build all packages:

catkin config --cmake-args -DCMAKE_BUILD_TYPE=Release
catkin build

Launch Simulator

Launch the gazebo simulation:

cd /home/XXX/catkin_ws/src/gazebo_turtlebot_simulator/launch/ 
roslaunch ./gazebo_closeloop_turtlebot.launch

By default the gazebo GUI is off. To enable gazebo GUI during simulation, simply set arg gui in gazebo_closeloop_turtlebot.launch to true:

<arg name="gui" default="true"/>

The IMU spec can be adjusted as well. The two example IMU specs provided are mpu_6000 and ADIS_16448. Switch IMU spec by editing arg imu_sensor:

<arg name="imu_sensor"  value="mpu_6000" />

Or

<arg name="imu_sensor"  value="ADIS_16448" />

Lastly, the visual sensor can be adjusted with arg 3d_sensor:

<arg name="3d_sensor"   value="fisheye_stereo" />

Or

<arg name="3d_sensor"   value="kinect" />

Or

<arg name="3d_sensor"   value="asus_xtion_pro" />

Launch Closed-loop Evaluation

Adjust the parameters in batch evaluation script, e.g. feedback_GFGG_stereo_batch_eval.py. Detailed descriptions on each parameter are provided.

After settting the parameters, start batch evalution:

cd /home/XXX/catkin_ws/src/gazebo_turtlebot_simulator/script 
python feedback_GFGG_stereo_batch_eval.py

An rviz config is provided for visualization:

rviz -d /home/XXX/catkin_ws/src/gazebo_turtlebot_simulator/closeloop_viz.rviz	

Results Collection

The closed-loop navigation output are recorded as rosbag.
To convert these rosbags to text files, clone the repo:

https://github.com/ivalab/mat_from_rosbag

Then edit and execute the batch conversion script, so that CLOSEDLOOP_DIR is the folder all closed-loop evaluation rosbags are saved within:

./script/batch_closedloop.sh

Finally, use the closed-loop evaluation script in SLAM Evaluation repo:

https://github.com/YipuZhao/SLAM_Evaluation

Refer to the evaluation script on how to quantify the navigation performance and latency consumption:

closeLoop_error_TRO21.m

Contact information

• Yipu Zhao [email protected]
• Justin S. Smith [email protected]
• Patricio A. Vela [email protected]