Composable Safety for Cooperative Robots

CoSMo COVR AWARD: Composable Safety for CoBots

CoSMo is an EU-funded COVR AWARD project with the focus on composable safety models for cooperative and collaborative robot applications. The project is carried out in collaboration with Fraunhofer IFF Magdeburg and Weidmüller Interface GmbH.

Goal of the project

The main target of this project is to enable system integrators to efficiently perform safety validations of potentially large numbers of cobot applications used in cooperative assembly processes. Nowadays, a cooperative robot workstation that facilitates a flexible arrangement of modular physical components must be individually certified upon each change of the setup and assembly process, which is in contrast with the required flexibility of agile manufacturing. 

In the course of the project, we are investigating whether the inherent modularity of the workstations can be lifted to the certification process. For this, three exemplary assembly processes are investigated and set-up in a modular cobot workstation. Their safety requirements are then analyzed, and a risk assessment is carried out for each individual workstation configuration. Based on the concept of composable safety models, an extended validation process for modular workstations will be developed in interaction with the COVR experts and fed back to the toolkit, which shall ease the certification of modular cobot applications.

A modular workstation that is considered as evaluation application is composed of different interchangeable workspaces that itself consist of stations. The chosen set of stations determines the products that can be manufactured with a particular configuration in the design space.

To reduce the effort of a risk analysis for modular workstations, a specific process is introduced in this project. The complexity of the “real” world is reduced by a dimension reduction into an abstract design space. In that space, risks are analyzed and risk reduction measures are applied in an iterative process to optimize the design configuration until only acceptable risks exists. Finally, the “real” world design is adapted to the optimized abstract design.


July 2019 - March 2020


  • Dennis Wigand
  • Dr. Sebastian Wrede