DARKO

Dynamic & Agile Production Robots.

 

Concept drawing of the project's content
Contact

Kai O. Arras

Robert Bosch GmbH

Kai Arras is heading the Robotics Research Program at Bosch, Chief Expert Robotics and Bosch PI of DARKO.

DARKO

Dynamic Agile Production Robots That Learn and Optimize Knowledge and Operations 

DARKO is an EU-funded research and innovation project on efficient and safe mobile robots for logistics in agile production. 

Safe, capable and efficient robots for factory logistics

In order to better adapt to changing conditions and market needs, the manufacturing industry worldwide is pushing towards more agile and efficient production. Agile production crucially depends on the effectivity of intralogistics — simply, the act of moving bits and pieces around the factory. Factory logistics robots have the potential to be a game changer, but only if they have the right capabilities. They need to be highly flexible, cost- and energy-efficient, and able to safely and smoothly operate in work environments shared with humans.

The DARKO project therefore sets out to realize a new generation of agile production robots, aiming to close the "skills gap" where today's robots fall short. We develop energy-efficient elastic arms that can pick items and throw them in a more human-like fashion. Importantly, we also research ways to make the robots able to operate safely within unknown, changing environments; in part by being aware of humans and their intentions to interact with them smoothly and intuitively. Our robots must also be easy to deploy in new factories to reduce the cost and effort currently required for mapping out a new site, devising routes and traffic rules, etc. Finally, DARKO's robots will have predictive planning capabilities, to decide for the most efficient action while limiting risks.

DARKO's Objectives

Our main overarching objective is to research and innovate for efficient and safe intralogistics robots in agile production. More concretely, we aim to realize the fundamental enabling technology that will make intralogistics in agile production flexible, robust and easily deployable. Our work is focused along the following five themes:

  • Efficiency in manipulation: Today‘s industry robots can pick objects but mostly in a static and stationary setup. DARKO will make possible more advanced manipulation by means of novel hardware that is safe and energy-efficient, better computer vision for seeing difficult objects, and agile control and perception for throwing.
  • Efficiency in human-robot coproduction: For intralogistics robots to be efficient, they must co-exist with human staff. We will increase efficiency and safety by learning activity patterns so that the robots can adapt to them, and by predicting people’s intents, as well as novel ways of communicating the robots‘ intents.
  • Efficient deployment: One of the main barriers to market of robots today is deployment effort. We will make deployment easier from the point of view of failure-resilient mapping and localization, as well as “auto-completion” of robot maps with existing maps of different type.
  • Risk-aware operation for safety and efficiency: We include risk assessment as a driving principle for the decisions taken by the robots, including the overall orchestrated behavior of robots, humans, and machine.
  • Demonstrate feasibility in an integrated system: Finally, a central theme is to demonstrate in practice, in an integrated system, how the scientific objectives above can be put to use. ARENA2036 is the perfect environment for this, where we will construct a replica of a use-case based on real-world requirements, to be used for demonstrating and validating our scientific results.

DARKO's Project Structure

The DARKO consortium brings together a unique combination of partners with highly complementary, world-leading expertise to tackle the application needs for robots that operate in intralogistics applications of agile production in environments shared with humans. A number of key technologies will be researched and further developed during the project within seven scientific work packages:

  • Novel elastic manipulators and flexible end effectors: Project partner Technical University of Munich (TUM), building on world-renowned expertise in safe manipulation, leads the work on design and control of a novel and efficient elastic manipulator in work package WP1. Project partner University of Pisa has a strong record in manipulation, and in particular end-effector design. They will contribute to the design and control of novel flexible end-effectors.
  • 3D perception and scene understanding: As part of WP2 (3D perception and scene understanding), partner Bosch will lead research that is conducted together with Örebro University, EPFL, University of Pisa and University of Lincoln on more robust, efficient and intelligent AI-based methods that will provide mobile and industrial robots with a better semantic understanding of their static and dynamic surroundings. To this end, they will combine multi-modal data from lidar and vision-based sensors using supervised, weakly supervised and self-supervised deep learning techniques.
  • Mapping and localization: Coordinating partner Örebro University will lead research on mapping and localization technology to further improve the robustness of existing SLAM systems and enrich them with higher-level information.
  • Safe dynamic manipulation: University of Pisa leads WP4, which covers research on novel methods for safe and dynamic manipulation. Project partner École Polytechnique Fédérale de Lausanne (EPFL), with their strong expertise in robotic throwing, will contribute to manipulation and throwing tasks in WP4. Technical University of Munich will focus on safety-related aspects, whereas Örebro University contributes with the connection of vision and grasping.
  • Human-robot spatial interaction: University of Lincoln, with its Lincoln Centre for Autonomous Systems (L-CAS), together with Örebro University and Bosch will research more capable approaches for social navigation in the presence of humans as part of work package WP5.
  • Predictive motion planning: Bosch, based upon their wide expertise in social navigation, leads WP6 (Predictive Motion Planning), which aims to investigate risk-aware and safe methods for planning and control of wheeled mobile robots in dynamic environments by pro-actively considering predicted future human states. Major contributors to this task will be University of Pisa and TU Munich with their existing background on safe and risk-aware planning.
  • Risk management: Project partner ACT Operations Research IT leads the technical WP7 (risk and scheduling), where they, together with University of Pisa, develop a risk assessment and forecasting framework to ensure both efficient and safe operation in human-robot shared environments.

DARKO has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 101017274.