Real-Time ROS (Robot Operating System) Analysis and Monitoring: Keeping Robots in Step

– Guest posting by Michael Stiller, Research Associate, Fraunhofer ESK

Michael Stiller Fraunhofer ESK
Michael Stiller, Fraunhofer ESK

Manufacturing robots have to operate reliably and safely. Fraunhofer ESK is working on ways to analyze and monitor open source robot operating systems in real-time in order to optimize production system robots.

Robots serve lots of functions: searching luggage for explosives, or welding and assembling in production halls. Some even play soccer against each other. In theory though, these robots are nothing more than a system of distributed software components that communicate with one another. Large robot systems consist of dozens of these software components, which are also referred to as communications nodes. Analyzing such systems when problems occur can be an extremely complex and time-consuming task.

Real-time analysis with DANA

This also applies to robots that function with open source robot operating systems (ROS). While ROS’ have onboard features that permit analysis of all of the message exchanges with the help of log files (offline error analysis), or stimulation of the application by repeatedly replaying the messages, real-time analysis is possible only to a limited extent.

In these situations Fraunhofer ESK’s DANA analysis framework can help. This framework offers the opportunity to compare the actual behavior of the ROS-based application to a model-based target behavior. To do this Fraunhofer ESK developed a DANA ROS node that forwards the ROS messages to the analysis framework where a target/actual comparison is carried out. Communications between the DANA node and the DANA framework was implemented with MQTT (message queue telemetry transport), an open-source message protocol for machine-to-machine communication.

DANA analysis framework Fraunhofer ESK (background: robot arm Autotrace)
DANA analysis framework by Fraunhofer ESK

Self-learning system

At the automatica trade fair, Fraunhofer ESK will have a demonstrator that simulates a pick & place scenario using robots and a turntable. The demonstrator replicates mechanical and software changes that impact the ROS application and visualizes the changes with the help of the DANA framework. The solution takes it one step further however, Apart from analyzing the ROS application, it also allows DANA to learn the behavior of the ROS application. For example: if changes to an existing, functioning ROS application are planned, DANA can learn the target behavior prior to the changes by automatically creating a model of the behavior. Once the changes are implemented, DANA monitors the application to determine if the current behavior is being violated.

DANA is not restricted to ROS applications. Practically any distributed application, such as those found in proprietary software architectures from various manufacturers, can be efficiently analyzed in real-time. That includes for complex industrial systems, high-grade connected IoT applications or autonomous vehicles.

 


Visit the IT2Industry Forum lecture on ‘So that robots don’t break ranks: Real-time analysis and monitoring for ROS (Robot Operating System)‘ (original title:
Damit Roboter nicht aus der Reihe tanzen: Echtzeitanalyse und Monitoring für ROS (Robot Operating System)‘) by Michael Stiller, Wissenschaftlicher Mitarbeiter, Fraunhofer Institute for Embedded Systems and Communication Technologies (Fraunhofer ESK), on Thursday, 19 June, 2018, 3:20 pm, at automatica 2018, hall B4.

Discover latest from robots and automation at automatica 2018 – plus industrial software and control systems for IIoT and Industry 4.0 at its subject area IT2Industry – 19 to 22 June, Messe München.
– Register for automatica 2018 ticket.

,,,,,,,,,

Leave a comment

Your email address will not be published. Required fields are marked *

Accept and Proceed

Cookies help us to improve our services. By using our website, you agree to our use of cookies. Learn more