Enhancing Swift and Socially-Aware Navigation with Continuous Spatial-Temporal Routing

Algorithm demo & Guidelines for Simulating Mobile Robots in Crowded Environment

• Click to view the project demonstration video: https://www.youtube.com/watch?v=wsg-De5rDfc&t=54s

This repository contains following items for the spatial-temporal routing project:

• MATLAB demo for the spatial-temporal routing algorithm
• Guidelines on how to implement the high-level path planning algorithm into a 3D pedestrian simulator using ROS packages
• A custom differential drive robotic wheelchair model in URDF format
• XML-based scene (generated from THÖR datasets: real human trajectories for human-robot interaction testing)
• MAT-based scene (artificially generated for testing robot navigation in large crowds)

Software Requirements:

System:

• Ubuntu 16.04.7
• [ROS Kinetic] (http://wiki.ros.org/kinetic/Installation/Ubuntu) (may work with other versions, tested and deployed on Kinetic)

Require Packages:

• ROS Packages: https://github.com/srl-freiburg/pedsim_ros
• Matlab Packages: https://web.casadi.org/get/

Installation

Step 1

Set Up the Testing Environment in Your Workspace (SPACiS)

git clone https://github.com/maprdhm/Spaciss.git  
cd Spaciss
git submodule update --init --recursive
cd ../..
catkin_make or catkin build (twice at the first time)

Step 2

• Download the robot_wheelchair file into your workspace (catkin_ws/src)
• Replace the experimental_package file from the Pedestrian_simulator file

Example usage

• launch the robot in the crowded environment: roslaunch experimental_package Scenario_test.launch

• To change the environment in the launch file: <arg name="scene_file" value="$(find experimental_package)scenarios/business_area/low/thormap.xml"/>

THÖR scenario: (Pedestrian_simulator/experimental_package/scenarios/business_area/low/thormap.xml)

Data communication

• The position of simulated pedestrians can be extracted via the ROS topic: /pedsim_visualizer/tracked_persons (not a regular pose message; requires custom processing in MATLAB: MATLAB ROS Custom Messages)

• Alternatively, the position of simulated pedestrians can be extracted using the top-view camera in RVIZ:

  Set the color of the crowd to blue in the top-view camera Record the video in MP4 format Use Support\Crowd(blue)_tracking_fixed_frame.m to track the crowd's motion

• 'MotionControl_ros_robot.m' is a motion controller that, given the reference path, continuously publishes cmd_vel commands to control the robot and follow the path.

Cite this work

• The detail of this work can be found in our paper: Enhancing Swift and Socially-Aware Navigation with Continuous Spatial-Temporal Routing