ROS_HRI

Python scripts with ROS integration to perform Human robot interaction:

• Head pose estimation
• Skeleton pose estimation
• 2D gestures recognition

Work in progress

Dependencies

Python3 with ROS (follow this to setup the environment https://medium.com/@beta_b0t/how-to-setup-ros-with-python-3-44a69ca36674).

Install the requirements of the project:

pip install -r requirements.txt

Docker installation

Clone the repository

git clone https://github.com/gonzalezJohnas/ROS_HRI.git

Build the docker image for the ROS_HRI

docker build - < Dockerfile

Then you can run the Docker image with --network host to used the host network and with -it to get a bash to run the build the application.

docker run --network host -it  ros_hri_gpu /bin/bash

All the code and dependencies are installed there just need to run the ROS_HRI application

root@1ec54b63ea34:/catkin_ws# source devel/setup.bash
root@1ec54b63ea34:/catkin_ws# python3 src/ROS_HRI/src/main.py <name_camera_node> 
				<pose_estimation_model_path>
				<hand_classifier_model_path> 
				<hand_gestures_label_file> 

Example of the command

 python3 src/ROS_HRI/src/main.py /camera/color/image_raw 
 /home/icub/catkin_build_ws/src/ROS_HRI/src/humanpose/checkpoint/checkpoint_iter_370000.pth
 /home/icub/Documents/Jonas/HandPose/cnn/models/hand_poses_wGarbage_10.h5 
 /home/icub/catkin_build_ws/src/ROS_HRI/src/handGesture/poses.txt