Rehab Robot Person Location & Time Tracking

Overview

This package contains nodes which track the person location & the time spent in the living lab map area. This package has been tested with the actual robot running navigation and person tracking AI model. Also, the same package has been tested in simulation which is described here. Also, this package contains a node which publishes the text labels to be overlayed on the living lab map in RVIZ. The below described information is provided according to the testing done on actual robot running navigation & AI models.

Pre-Requisites

1. This package is tested with ROS Melodic so preferably use the package with same ROS version.
2. This package depends highly on the navigation stack & AI model which finds the person in the camera frame and publishes its location on a ROS topic. Follow these instructions for setting up the AI model on the robot, moreover for running the navigation on the robot, follow instructions here.

Setup

After following the step 2 in the pre-requisites, you must have the following in working:

• Robot must be running navigation. Verify that the topic /amcl_pose is being published by the navigation stack & the robot is able to navigate by accepting goal from RVIZ.
• The AI model is running correctly and the person location is being published on rostopic /person_loc in camera frame.
• Make sure that the complete transform of the robot is being published correctly on /tf topic.

Usage

After following the above setup process, run the following nodes:

1. Run the following launch file to display the Location labels as RVIZ markers as shown below in the image,

   roslaunch rehab_person_loc rehab_person_loc_pubMarkers.launch

   

2. Run the following main launch file,

   rosrun rehab_person_loc rehab_person_loc_time.launch

   Subsribed Topics:

   1. amcl_pose (geometry_msgs/PoseWithCovarianceStamped)

      Pose of the robot in the MAP frame.

   2. person_loc (geometry_msgs/PointStamped)

      Location of person in camera frame.

   Published Topics

   1. person_loc_estimated (geometry_msgs/PointStamped)

      Computed location of person in MAP frame

   2. location_tag(rehab_person_loc/location_info)

      Location of the robot and person in MAP.

   3. time_info(rehab_person_loc/time_info)

      The duration the person spents @ various locations in the tagged area in MAP.

   Messages

   1. location_tag.msg

   time stamp
   string frame_id
   string robot_location
   string person_location

   2. time_info.msg

   int32 kitchen_time
   int32 lounge_time
   int32 entrance_time
   int32 lobby_time
   int32 tvRoom_time
   int32 bedRoom_time
   int32 away_time

Output

Check the below two published topics for the output shown below:

• The topic location_tag will show the person and robot location in the MAP frame. Ouput will be like the one shown below.

  

• The topic time_info will show the duration (in seconds) the person spends at various locations inside the MAP. Ouput will be like the one shown below.

  

• Add the topic person_loc_estimated to RVIZ for visualizing the location of the person inside the MAP. The output will be like the one shown below (person location shown as purple colored marker on the MAP)

  

• The package also logs the person and robot location data along with the system clock time stamps in a text file. The log file is stored in rehab_person_loc/logs folder with the name rehab_person_loc.txt. Sample file can be checked here

---

Contact

Author: Ali Ahmad
Designation: Senior Robotics Engineer
Email: [email protected]