Human Pose Demonstration/Following for Baxter Robot

This project achieves how to demonstrate a Baxter robot (enable the Baxter robot to follow human actions) using human pose estimation.

Environment

• System: Ubuntu 18.04
• Sensor: Kinect v2
• Robot: Baxter pkg

Installation Steps

1. Kinect v2 Installation

Follow the instructions provided in the link below to set up Kinect v2:

• Kinect v2 Installation Guide

2. kinect2_tracker Installation

Install the kinect2_tracker package for human pose tracking:

• kinect2_tracker Installation Guide

Note: Use NiTE-Linux-x64-2.2 instead of NiTE-Linux-x64-2.0.

3. Baxter Robot Setup

Set up the Baxter robot by following the instructions below:

• Baxter Installation Guide
• Source Code: Rethink Robotics Baxter GitHub Repository

4. Kinect-Based Arm Tracking

Clone and set up the kinect_based_arm_tracking package:

• Source Code: kinect_based_arm_tracking GitHub Repository

Details of the code

check the README.md in each sub folder in src/ for details

Running the Demo

To run the demo, open multiple terminals and execute the following commands:

Terminal 1: Start Baxter Robot

./baxter.sh
rosrun baxter_tools enable_robot.py -e # enable robot at the start

Terminal 2: Launch Kinect2 Tracker

roslaunch kinect2_tracker tracker.launch

Note: It may take time to model the person. You may need to retry this step multiple times until the "tf" of the person is successfully published.

Terminal 3: Publish Human Pose Data

cd ~/ros_ws/src/kinect_based_arm_tracking/scripts
python tf_listen_v8_puber.py

Terminal 4: Control Baxter Robot

cd  ~/ros_ws/src/kinect_based_arm_tracking/scripts
python tf_listen_v8_controller.py

References

https://blog.csdn.net/lwq123free/article/details/96335957