A typical application issue that today’s robotics research is faced with is bin picking. Picking in unfamiliar environments and the autonomous response to changing circumstances remain a major challenge for many industry applications. The successful contract logistician FIEGE facilitated a use case to trial the SAINT project, a programme that is sponsored by the Bavarian Research Foundation. Munich-based robotics specialist FRANKA EMIKA GmbH and two chairs from the Technical University of Munich (TUM) oversaw the project’s technical implementation.
Many industries – the automotive industry being one example – depend on a multitude of production processes for which robotics are indispensable. According to the 2020 World Robotics Report published by the International Federation of Robotics (IFR) in September of last year, factories around the world employ somewhere over 2.7 million robotic systems. Most of them are regular palletting, cutting, or welding robots which perform their pre-programmed workflows completely autonomously.
Service robotics in particular has made huge strides these past years. The respective solutions are being applied more and more frequently in the industry nowadays, to enable a closer collaboration between man and machine. So-called cobots, or collaborative robots are intended to interact directly with humans to solve more complex tasks. They no longer operate behind protective screens but with the help of cutting-edge sensors and camera technology that detect obstacles to prevent collisions.
Bin picking in fashion logistics
Even within logistics, automation technology and robotic assistance are now gaining ground. The goals here include performance increases and reducing the workload of employees while cutting costs. The issue of item picking is one of the driving forces behind the research.
Bin picking refers to the robot-guided picking of items that are placed randomly in a bin. “Many areas, especially when separating supple items, are still dominated by a major discrepancy between the respective solutions that research devises and authentic industry applications”, Jonas Wittmann from the Chair of Applied Mechanics of the Technical University of Munich says. Together with his colleagues from the TUM and employees of FRANKA EMIKA GmbH, a robotics specialist, he has been working on this subject under the SAINT research project these past two years.
FIEGE also came on board as a partner to this project by facilitating a use case at its Mönchengladbach branch from the field of fashion logistics. The case in point tackled the automatic repacking of various items of clothing from boxes onto a split tray sorter. Currently, this step is still being handled manually by employees. Rolf Beckmann, Director Engineering at FIEGE, explains: “The goal was to realise a productive application of robotics which would require relatively few adjustments to the existing situation and help man with their job.”
Plastic bags as the ultimate challenge
Initially, the scientists around Wittmann sought to programme the software for the controls: “We implemented five different modules, all of which were based on freeware. Some were responsible for calibrating the robot’s arm, its motion planning and its autonomous troubleshooting.” For this, a camera system estimates a variety of possible picking poses which the system optimises and the robot then executes.
Alongside this, the robotics experts of FRANKA EMIKA had to tackle a practical issue, as Lukas Hausberger, Robotics & Machine Learning Engineer, reported: “We pursued two approaches to picking garments wrapped in plastic bags, one of which we quickly dismissed.” Attempts using a vacuum picking tool showed that the suction created was insufficient to perform swift movements. To solve this, they borrowed an idea from the ancient Greeks, amongst other things: “We came up with a picking robot that has two parallel grippers equipped with tactile and optical sensor technology. This facilitates force-sensitive gripping without damaging fragile objects. In combination with the Archimedes finger, which is based on the namesake Archimedean spiral, we also devised special gripper jaws. The additional rotation creates friction and pulls the plastic bags between the gripper’s fingers”, Hausperger explains.
Man and machine learning hand in hand
A further vital parameter of the project was troubleshooting. Initially, this was to run autonomously or in part autonomously using FDIR (Fault Detection, Isolation and Recovery) algorithms taken from aerospace technology, as Wittmann explains: “The goal was for the picking arm to autonomously detect faults, assess possible consequences, and calculate a respective alternative approach.”
For this, the robot’s arm has a remote control. “Teleoperations allow the incorporation of man’s expertise. In consequence, we can expedite our response to defects and facilitate the hybrid automation of complex process flows”, the researcher says. The picking arm’s motion sequences were thus increasingly optimised in the course of the project.
Practical test demonstrates potential
At the beginning of the year, two on-site tests took place at the FIEGE location in Mönchengladbach, which is where the fashion logistics specialist oversees operations for an international fashion company. After a pre-trial in January, the final demonstration took place at the end of February following the installation and acceptance of the robotic arm. For the Proof of Concept, the picking arm had to repack multiple different garments from a bin into the split tray sorter. “The system was not yet sufficiently developed to estimate the weight of the objects. The failure rate for jackets was clearly higher than for t-shirts”, Wittmann says.
However, Wittmann was still satisfied: “We solved the task and developed a robust and low-budget robotic application for picking in unfamiliar environments.” But he admits that there is still room for improvement. The robotic arm’s productivity, for example, is currently still behind that of the employees.
At the beginning of the year, two on-site tests took place at the FIEGE location in Mönchengladbach.
“Businesses must become innovation drivers”
From FIEGE’s perspective, the initial conclusion was equally a positive one, even if dichotomous. Rolf Beckmann: “As such, the SAINT project was a success. We are impressed by the performance of the entire team.” As a logistics company, FIEGE is predominantly busy with the question as to when collaborative robotics and bin picking can be applied economically in fashion logistics. “There is still a distance to cover in that regard”, the Director adds.
Yet Beckmann also sees his own responsibility because more often than not, the interests between developers and users differ. “As an innovation-driven company, we see ourselves as a partner in research and aim to impart our expertise even in the future.”