Handling with AI-enhanced Robotic Technologies for flexible ManUfacturing

HARTU develops components for the automatic planning and control of gripping and assembly processes in robotics, as well as innovative gripping concepts based on electroroadhesion. The goal is to make industrial manufacturing more efficient, flexible and reconfigurable. The main technological goals are self-supervised identification and control of gripping strategies, learning and control of robotic assembly techniques, AI-based multimodal perception for visual control and monitoring of handling operations, and development of versatile soft grippers with electro-active fingertips. The technologies will be evaluated in 5 different industrial scenarios: Kitting, Assembly, Sorting, Packaging, and machine tending.

Duration: 01.01.2023 till 31.12.2025
Donee: German Research Center for Artificial Intelligence GmbH
Sponsor: European Union
Grant number: 101092100

Fundacion Tekniker (ESP), AIMEN Centro Tecnológico (ESP), Industrial Technology Research Institute (TWN), Politecnico di Bari (ITA), Omnigrasp (ITA), Engineering Ingegneria Informatica (ITA), FMI Drachten B.V. (NLD), ULMA Manutención S. Coop (ESP), Tofaş Türk Otomobil Fabrikası (TUR), Philips Consumer Lifestyle BV (NLD), Tecnoalimenti SCPA (ITA), INFAR Industrial Co., Ltd. (TWN), Deep Blue SRL (ITA)

Application Field: Logistics, Production and Consumer
Related Projects: Hybr‐iT
Hybrid and intelligent human-robot collaboration – Hybrid teams in versatile cyber-physical production environments (11.2016- 10.2019)
Flexible Interaction for infrastructures establishment by means of teleoperation and direct collaboration; transfer into industry 4.0 (07.2017- 12.2021)
Multipurpose robotics for mAniPulation of defoRmable materIaLs in manufacturing processes (04.2020- 07.2023)
Virtuelle Sektionsmontage und Systeminstallation (06.2021- 08.2024)
Intelligent Human-Robot Collaboration (03.2015- 06.2016)
Related Robots: OmniPick
iMRK Dual-Arm Robot (Photo: Annemarie Popp, DFKI GmbH)
Related Software: ARC-OPT
Adaptive Robot Control using Optimization
Hybrid Robot Dynamics
Behavior Optimization and Learning for Robots

Project details

Overview of the software components developed by DFKI in HARTU (Dennis Mronga, DFKI)
Example of manual operations in manufacturing, here assembly and packaging (Photo: HARTU Project)

The handling of parts in manufacturing environments is a concatenation of some of these basic actions: part grasping, manipulation, transportation, and release. In highly automated lines, each of these actions is done by special machines or by means of robots that are programmed ad-hoc. Grasping devices, using several working principles (prehension, vacuum, magnetism, etc.) are selected for each part or family of similar parts, introducing tool exchangers when required. Reconfiguration of these lines has a high economic cost and is time consuming.

Moreover, when flexibility is required due to the variability of products to be handled, a highly unstructured environment, or the very delicate nature of products this approach is not feasible. In those cases, manual handling is the adopted solution.

For this reason, HARTU provides components for automatic planning and control of grasping, release and contact assembly tasks, and proposes innovative gripping concepts based on electroadhesion for the handling of many different products. These components are integrated through a reference architecture and supported by perception capabilities and application development support tools, with the overall goal of making manufacturing lines more efficient, flexible, and reconfigurable.

In HARTU, DFKI is primarily concerned with learning and controlling robotic assembly skills that involve a variety of different contact situations. Therefore, motion sequences and contact wrenches shall be learned from human demonstrations and generalized with respect to novel situations. The acquired skills will be generic; the robot control parameters shall adapt automatically to the current contact situation.

The developed methods will be demonstrated in the mass production of consumer goods, in this case electric razors, at Philips Consumer Lifestyle B.V.. They are intended to make assembly lines more flexible and easier to reconfigure.



Gaussian Mixture Likelihood-based Adaptive MPC for Interactive Mobile Manipulators
Dimitrios Rakovitis, Dennis Mronga
In 2024 IEEE International Conference on Robotics and Automation (accepted for publication), (ICRA-2024), 13.3.-17.3.2024, Yokohama, IEEE, 2024.

Back to the list of projects
last updated 04.01.2024
to top