dratasich / shsa_ros

ROS wrapper for SHSA

ROS Wrapper for SHSA

Self-Healing by Structural Adaptation (SHSA) for the Robot Operating System (ROS).

This repo implements the ROS interface for the shsa library and shsa-problog library, to demonstrate SHSA on the mobile robot Daisy.

Dependencies

• The ROS nodes shsa_node.py, transfer_function_node.py, watchdog_node.py depend on the shsa library.

• The ROS node monitor_node.py depends on the shsa-problog library.

Note that the libraries cannot be both in PYTHONPATH (name clash). We used custom docker images to run the ROS nodes.

Run

Execute SHSA node:

$ rosrun shsa_ros shsa_node.py _model:=/path/to/shsa-model.yaml &

Substitute a topic (send a goal, i.e., trigger an action of the SHSA node):

$ rostopic pub -1 /substitute/goal shsa_ros/SubstituteActionGoal

Trigger auto-completion by pressing tab two times, the content of the goal message can be filled (goal.topic).

Demo

This package also includes application launch files, e.g., to demonstrate SHSA on the mobile robot Daisy.

To run the demo you will need additional ROS and python packages.

Collision Avoidance with Daisy - Breakdown of the Laser

Get the shsa library and install the python packages needed (see the requirements.txt file). Add the library to PYTHONPATH. Source the catkin workspace containing general-ros-modules and shsa_ros.

Run the application:

$ roslaunch shsa_ros demo_daisy.launch

which starts the drivers and the teleoperation node of Daisy. Use the keys w|a|s|d and space to move and stop the robot. In case the motors haven't been enabled (a start after p2os_driver is necessary), launch enablemotors.launch.

Start the watchdog and self-healing engine with:

$ roslaunch shsa_ros demo_breakdown.launch

When the laser breaks (e.g., switch off the power supply), the watchdog_node triggers the shsa_node to substitute the minimum-distance-to-an-obstacle (dmin) calculation.

Collision Avoidance with Daisy - Spoofing Attack

Get the shsa library and shsa-problog library and install the python packages needed (see the requirements.txt file). Add the proper library to PYTHONPATH (depends on the nodes to start). Source the catkin workspace containing general-ros-modules and shsa_ros.

Run the application:

$ roslaunch shsa_ros demo_daisy.launch

Start the self-healing engine with:

$ roslaunch shsa_ros shsa.launch

Start the monitoring exploiting redundancy:

$ roslaunch shsa_ros monitor.launch

Plot minimum-distance-to-an-obstacle dmin:

$ rqt_plot /emergency_stop/dmin/data /dmin_monitor/value_0/data /dmin_monitor/value_1/data

When an attacker spoofs wrong laser data to the ROS network, e.g., by:

$ rostopic pub /hokuyo/scan sensor_msgs/LaserScan "ranges: [3.0]" -r 100

the monitor_node triggers the shsa_node to substitute the minimum-distance-to-an-obstacle (dmin) calculation.