Crowd-Aware-Navigation-Based-on-A-Risk-Map-and-Multi-Object-Optimization-Approach

1. Introduction

This repository includes codes employed in the paper Crowd-Aware-Navigation-Based-on-A-Risk-Map-and-Multi-Object-Optimization-Approach.

NOTE: To properly use the code, please modify the path in dangermap_core.py to your own path.

2. File Structure

• dangermap_core: Contains the code for danger map construction and action generation.
• world and model: Contains the world and model file needed for gazebo simulation.

3. Installation

ROS2

Any ROS2 version should be fine, this code is tested under ROS2 foxy and galactic.

Turtlebot

Install Turtlebot3 packages following this tutorial:

https://emanual.robotis.com/docs/en/platform/turtlebot3/quick-start/

Remember to install DynamixelSDK, turtlebot3_msgs, turtlebot3, and turtlebot3_simulations from source. And copy the 3_cylinders.model inside world and model fold to the gazebo world folder. cylinder_model need to be pasted to ~/.gazebo/.

Compile and Run

cd /your/ws/

colcon build May face some dependency error, install them with:

rosdep install --from-paths src -y --ignore-src

Python-RVO2 need manual installation.

After compiling, to launch gazebo:

source install/setup.bash ros2 launch turtlebot3_gazebo turtlebot3_world.launch.py

To run our code:

ros2 run dangermap_core dangermap_core --policy ca_lidar

This command will launch the quantitive experiments where the vehicle is running in different configurations: radius:3-20, number of obstalce: 3-20, speed: 0.2-1. Each setup will be tested for 100 times. Success and fail count as well as the mean computation time will be printed to the terminal.