Sebastian ZarnackFraunhofer IIS/EAS
Deployment of adaptive robotic systems with dynamic collision avoidance
While conventional industrial robots have provided an essential prerequisite for the automation of manual, heavy and monotonous work in industrial manufacturing over the last decades, most of today's robotic systems still lack a comprehensive sense of their environment, their own capabilities and a comprehensive contextual understanding of the surrounding processes and resources.
Thus, the potential flexibility of a robotic system cannot fully be exploited due to a lack of perception and information about the outside world. The capabilities to collaborate with other robot systems are limited and mostly only realizable via central control.
However, these capabilities will play an increasing role as success factors in the factory of the future, in which increasingly smaller batch sizes, a drastically increased variety of variants, and also an ever faster and more flexible response to changing customer needs will be required.
Fraunhofer IIS/EAS is therefore researching flexible, adaptive systems in robotics for the factory of the future. While in conventional robotics trajectories are usually programmed in a fixed way, the ability to detect obstacles and to avoid them dynamically based on online trajectory generation at program runtime plays a key role for adaptive robot systems.
This requires the integration of external sensors and their evaluation in real time. From different data sources a comprehensive picture of the environment is generated by sensor fusion, which includes static as well as dynamic obstacles.
For this purpose, we want to show a demonstrator that is currently being developed as part of an ongoing diploma thesis. As an example, an active optical tracking system is used as an external environmental sensor. With the help of this system, the position and orientation of exemplarily selected obstacles can be dynamically imported into a simulation environment under ROS 2. This environment is used as a basis for hybrid motion planning using the well-known Moveit! framework for the Franka Emika robot. As a result, the autonomy and resilience of the system to unexpected disturbances and changes in the environment, as well as the safety of the interacting user, is significantly increased.