The EU H2020 project DARKO (2020–2021) set out to realize a new generation of agile production robots that have energy-efficient elastic actuators to execute highly dynamic motions; that are able to operate safely within unknown, changing environments; are easy (cost-efficient) to deploy; that have predictive planning capabilities to decide for most efficient actions while limiting associated risks; and which are aware of humans and their intentions to smoothly and intuitively interact with them. To maximise its impact, DARKO is aligned with use cases at the largest manufacturer of home appliances in Europe (BSH Home Appliances).
As leader of the work package on “3D Perception & Scene Understanding”, I oversaw five different tasks spanning from broader-level 3D scene understanding and real-time 3D perception of humans over perception for manipulation and in-hand grasp perception. One particular challenge in DARKO was to deploy such a complex and powerful perception stack on a resource-constrained mobile robot, equipped with only one desktop-grade GPU and an Nvidia Jetson Orin.
The project’s milestone demonstrations and stakeholder meetings took place at ARENA 2036 (University of Stuttgart, Germany) and at KI.Fabrik (Technical University Munich / Deutsches Museum). The final project results have also been demonstrated at the Automatica 2025 fair in Munich, as part of a larger AI.Society exhibition hosted by the Munich Institute of Robotics and Machine Intelligence (MIRMI).