Mobile Manipulation Demo

In this demo a mobile manipulator has to achieve a mobile pick&place task.

The robot is controlled both via a fault-tolerant Finite State Machine and a Behavior Tree.
The simplest version of the mobile manipulation task is realized also with a sequential design of a FSM.

Setup

This code was tested on Ubuntu 20.04 with ROS Noetic.

To make sure that dependencies are pulled, run the following command after cloning this repo into a ROS workspace:

git submodule update --init --recursive

System dependencies need to be installed by running:

./submodules/moma/install_dependencies.sh

Finally, the code can be built using:

catkin build mobile_manip_demo

Run Experiments

The type of task the mobile manipulator can perform, can be defined by the parameter experiment:

1: pick the cube2 and place it in the delivery station;
2: pick the cube2 and place it in the delivery station but go to recharge the batteries when they are low;
3: pick the cube2, place it in the delivery station, then dock the robot but go to recharge the batteries when they are low.

In a first terminal, launch the experiment environment in a gazebo simulation:

roslaunch mobile_manip_demo run_gazebo.launch

Then, the fault tolerant state machine can be launched through

rosrun mobile_manip_demo fault_tolerant_sm.py

the sequential state machine can be launched through

rosrun mobile_manip_demo sequential_sm.py

whereas the behavior tree can be launch through

rosrun mobile_manip_demo behavior_tree.py

Workspace overview

the control policies are defined in src
the robot skills are defined in scripts

Media

The video showing the mobile manipulator solving the task while controlled by the three policy representation is available at this link.

The images of the policy representations can be found in the imgs folder.

TODOs

Add instructions to install missing dependencies: