This repo contains all necessary package to have the Robot interract with its enviroment. The main focus is on the arm of the objects and focuses on mainly two objectives. Grasb and place objects given a position.
Additinally a package is included which positions the arm of the robot in certain predifined poses making sure the sensors of the robot are not blocked by itself.
Further deatails regarding each package can be found in its corresponding readme.
To start the manipulation servers run:
roslaunch suturo_manipulation_launch start_manipulation_servers.launch
You can run:
rosrun int_test_marker int_tes_marker.py
This will publish an interactive marker u can use to grasp, place at certain position. Additionaly you can select pick a pose the robot shall take.
- Install giskard https://github.com/SemRoCo/giskardpy
- (optional | lightweight simulator for testing) Install the iai_hsr_sim: https://github.com/code-iai/iai_hsr_sim
- Make sure you have suturo_resources (clone into src folder) https://github.com/SUTURO/suturo_resources
- clone the manipulation repo into src https://github.com/SUTURO/suturo_manipulation.git
How to start (detailed):
- source the workspace:
user@<computer>:<path-to-ws>$ source devel/setup.bash
- (optional | starts gazebo with an empty world) start gazebo:
roslaunch hsrb_gazebo_launch hsrb_empty_world.launch
- Make sure the simulation is started in gazebo by clicking the arrow in the bottom left control window
- Run the manipulation server (and their requirements):
roslaunch suturo_manipulation_launch start_manipulation_servers.launch
- (With iai_hsr_sim:)
roslaunch suturo_manipulation_launch start_manipulation_servers.launch sim:=true
Visual Servoing (Inspired by https://github.com/osrf/tensorflow_object_detector)
Make sure to install tensorflow
version 1.14.0. Also copy the modified hsrb_gazebo_bringup
package mentioned in the Projektdoku into the src
folder of your workspace. Due to legal reasons, it is not uploaded to any SUTURO repository.
To use the Visual Servoing, checkout the repositories of Giskard, Suturo_Manipulation and Suturo_Resources to the visual servoing branch. After that one can start the visual servoing by the provided launch file:
roslaunch suturo_bringup visual_servoing_bringup.launch
Developing for visual servoing requires a virtualenv
setup, because of the usage of python 3.6+. Follow the instructions in https://github.com/osrf/tensorflow_object_detector to setup the environment.
Follow the guides in suturo_manipulation/tensorflow_object_detector/src/object_detection/g3doc
to train new detection models.