Project adidas AIP, industry 4.0 with industrial robot

AIP@adidasAG

Feasibility study Automated Item Picking @ adidas AG

Context

Picking and placing articles is a central task in many logistics and production processes. In practice these tasks, known as item or bin picking , are still mostly carried out manually. This is very expensive for companies and a monotonous and tiring work for employees. Due to the constantly growing demand for e-commerce, the requirements on the efficiency of the associated logistics processes are increasing.
The use of Automated Item Picking (AIP) is a central component for overcoming these challenges of logistic processes. AIP systems basically combine two thematic areas: the item detection through a computer vision system and the item handling by a gripper equipped robot. While the handling of the items can usually be solved successfully using either a clamp or a suction gripper, a robust item detection often leads to great difficulties in chaotic and cluttered scenes.

Project goals

In a cooperation project between Karlsruhe University of Applied Sciences, adidas AG, KUKA Deutschland GmbH, Zimmer Group and Roboception GmbH, a prototypical solution is developed to automate the order picking process for sports and fashion articles. The primary goal is to detect and pick shoeboxes and plastic-wrapped apparel fully automated from a load carrier. The developed prototype should prove the general feasibility of the provided use case using a robotic solution. Furthermore, the goal is to produce results regarding different performance indicators such as the reachable cycle time and system reliability.

Publications

Publications project team

Weisenböhler M, Walz M, Wurll C: Automated Item Picking : Machine Approach to Recognize Sports and Fashion Articles. In: Christ A, Quint F (Hrsg.): Artificial Intelligence : from Research to Application. L'Intelligence Artificielle : de la Recherche á l'Applicacion = Künstliche Intelligenz : von der Forschung in die Anwendung (Offenburg, 13.03.2019), Karlsruhe: Karlsruhe University of Applied Sciences 2019, p. 71-75.- ISBN 978-3-9820756-0-0 (Electronic publication: https://arxiv.org/abs/1903.08495)
Weisenböhler M, Wurll C: Automated Item Picking for fashion articles using Deep Learning. In: ISR 2020; 52th International Symposium on Robotics. 52th International Symposium on Robotics (ISR 2020) (online, 09.-10.12.2020), Piscataway: IEEE 2020, p. 243-250.- ISBN 978-3-8007-5428-1 (Electronic publication: https://ieeexplore.ieee.org/document/9307478)

Student works

Augenstein, P: "Kameragestützte 3D Lage- und Positionsbestimmung zur Greifpunktplanung in der Intralogistik am Beispiel von Schuhkartons", bachelor thesis; Karlsruhe University of Applied Sciences, 31.08.2019
Braun, D: "Analyse und Evaluierung von Tiefensensoren für die Objekterkennung am Beispiel eines vollautomatisierten,robotergestützten Kommissionierprozesses", master thesis; Karlsruhe University of Applied Sciences, 28.02.2019
Deck, S, Bochinger, M: "Entwicklung und Evaluierung eines hybriden Greifsystems zur Automatisierung eines Kommissionierprozesses in der Intralogistik unter Einsatz numerischer Verfahren", joint Master thesis; Karlsruhe University of Applied Sciences, 28.02.2019
Gerlach, M: "Transferfähigkeit synthetischer Bilddaten für tiefe neuronale Netze", master thesis; Karlsruhe University of Applied Sciences, 27.02.2020
Kern, P: "Entwicklung eines Verfahrens zur Kollisionserkennung für das Automated Item Picking", bachelor thesis; Karlsruhe University of Applied Sciences, 29.02.2020
Schätzle, M: "Entwicklung eines vollautomatisierten, robotergestützten Kommissionierkonzepts für die Logistik – Einsatz klassischer Bildverarbeitungsverfahren zur Objekterkennung", master thesis; Karlsruhe University of Applied Sciences, 31.08.2018
Sobov, J: "Entwicklung, Implementierung und Erweiterung von Prozesslösungen zur Steigerung der Prozesssicherheit einer robotergestützten Kommissionierlösung in der Intralogistik", bachelor thesis; Karlsruhe University of Applied Sciences, 30.08.2019
Walz, M: "Entwicklung eines vollautomatisierten, robotergestützten Kommissionierkonzeptes für die Logistik – Einsatz von maschinellem Lernen zur Objekterkennung", master thesis; Karlsruhe University of Applied Sciences, 31.08.2018