ICRA 2020

Recreating navigation performance results in egoTEB: Ego-centric, Perception Space Navigation Using Timed-Elastic-Bands.

System Requirements and Dependencies

• Ubuntu 16.04
• ROS Kinetic
• catkin workspace

Note: If you encounter any problems, please create an issue.

Setting up Workspace

1. Install wstool and other system dependencies

sudo apt-get install python-rosdep  python-wstool  build-essential python-rosinstall-generator python-rosinstall

2. Navigate to the root of your catkin workspace

3. Initialize wstool workspace and download all packages

wstool init src https://raw.githubusercontent.com/ivalab/NavBench/master/ICRA2020/rosinstall 

4. Install any missing system dependencies:

rosdep install --from-paths src -i -y

5. Compile and source the workspace

catkin build
source devel/setup.sh

6. (Optional but recommended) Periodically check for updates to this demo:

wstool merge -t src https://raw.githubusercontent.com/ivalab/NavBench/master/ICRA2020/rosinstall
wstool update -t src

Running Benchmarks

Demo

To run a single navigation experiment of TEB navigating in a room populated with random obstacles, run the following command:

rosrun nav_scripts gazebo_master_demo.py

By default, the simulation gui is hidden. To display it, set the argument gui in navigation_test/configs/launch/gazebo_turtlebot_empty_room_20x20_world.launch to true. It is highly recommended to return it to false before running the full benchmark (below).

Run Navigation Benchmark

A benchmark comparing navigation performance of DWA, TEB, and egoTEB can be run as follows:

1. Create directory ~/simulation_data for logging test outcomes.

2. (Optional) Set the number of concurrent experiments to run using the num_instances variable at the beginning of the main function at the bottom of the navigation_test/scripts/scripts/gazebo_master.py file. For most systems, this should be left at 1. For reference, a 24 hardware thread Intel Xeon E5-2640x2 @ 2.50GHz (Single Core Passmark 1468, Multi-Threaded score of 14649) can generally run up to 3 concurrent experiments without issue. Overloading your system can negatively impact the experiments.

3. rosrun nav_scripts gazebo_master.py and wait for it to complete. If something goes wrong and the process does not exit cleanly, try rosrun nav_scripts killall.sh

4. Replace ~/simulation_data/results_... in navigation_test/scripts/scripts/analyze_results.py with the full path of your results file. rosrun nav_scripts analyze_results.py to generate a markdown table of results.

Timing Demo

1. Download sample rosbag of prerecorded data from: ''' https://drive.google.com/file/d/10glIc6hyeFE1OTMfC8wrL3jjvh4lklRz/view?usp=sharing '''

2. Replace [full path to]demo_rosbag_2020-05-30-17-05-50.bag in nav_scripts/scripts/launch/rosbag/rosbag.launch with the full path to the downloaded rosbag

3. Run timing demo

rosrun nav_scripts timing_demo.py