SamueleSandrini / skeleton_trajectory

Thesis project about human keypoints identification, 3d-reconstruction and filtering for collaborative robotics.

Skeleton Trajectory Repository

Welcome to the repository Project of CARI and Vis4Mechs Laboratory, University of Brescia, Italy !

This repository is a thesis project and aims at developing a ROS-based framework to identify human keypoints and to make the 3d-reconstruction of them inside a collaborative robotic cell, usefull for human movements tracking.

Start the skeleton node

The entire framework can be executed in a simplified manner for a user by means of a launcher file, which takes care of calling the various nodes:

roslaunch skeleton_trajectory skeleton_trajectory.launch

Launch parameters

The following parameter is available:

• kalmanBase: it specifies the filter typology.
  • kalmanBase:=True is the default condition and performs filtering with the points managed independently of each other (with assumption of constant acceleration in cartesian space).
  • kalmanBase:=False performed filtering with the model of the limb kinematics of person.

Requirements

• pyrealsense2:

pip install pyrealsense2

• Upgrade scipy: nedded for rotations

pip install --upgrade scipy

Maintainers

• Samuele Sandrini, SamueleSandrini
• Manuel Beschi, ManuelBeschi