RoboCup Logistics League: an interview with Alexander Ferrein, Until Hofmann and Wataru Uemura

The Industrial League enviornment at RoboCup2025.

RoboCup is a world scientific initiative with the purpose of advancing the state-of-the-art of clever robots, AI and automation. The annual RoboCup occasion passed off from 15-21 July in Salvador, Brazil. The Logistics League kinds a part of the Industrial League and is an application-driven league impressed by the commercial situation of a sensible manufacturing unit. Forward of the Brazil assembly, we spoke with three key members of the league to search out out extra. Alexander Ferrein is a RoboCup Trustee overseeing the Industrial League, and Until Hofmann and Wataru Uemura are Logistics League Govt Committee members.

May you begin by giving us an introduction to the Logistics League?

Alexander Ferrein: The concept of the Logistics League is to have robots serving to in intra-production logistics. The taking part in discipline is ready up with completely different machines and the robots must deliver uncooked supplies and merchandise to the machines and choose merchandise up from them. There are orders coming in for various merchandise of various complexities. The concept is that the robots ship these merchandise after they’ve been machined, at a sure handover level, after which the workforce can be awarded factors. The setup is that we now have six machines per workforce, and three robots working within the good manufacturing unit.

There are two groups competing on the identical time on completely different sides of the sphere. Many of the machines are on the house facet of the sphere, however some machines are additionally on the opponent’s facet of the sphere. The groups want to indicate primary robotics abilities like navigation abilities and collision avoidance. For the manufacturing of merchandise we now have little discs which have completely different colors and could be stacked on prime of one another, they usually have completely different caps. Now we have round 550 completely different gadgets that may be produced. We don’t focus a lot on the dealing with of the components, due to this fact the manipulating mechanisms are fairly easy and are often customized constructed by the groups. They simply must seize these discs and drive them round and put them on conveyor belts on the machines.

The primary focus is on the manufacturing logistics and the planning section. The complicated merchandise must be produced or machined by a workforce of robots – with out this planning and workforce work it wouldn’t be doable to ship the merchandise throughout the allotted time within the competitors.

Till now, we had been supported by Festo didactics, who provided the machines. Nonetheless, they pulled out in February and instructed us that they received’t help us in bringing the machines to Brazil. Our workforce in Aachen has an entire discipline arrange, so we’re within the (not really easy) means of packing the machines up in pallets and transport them to Brazil.

Until Hofmann: One essential element is that every one the merchandise that must be manufactured, the orders for these come on-line, and the variety of doable merchandise may be very excessive. Due to this fact, you may’t do any planning prematurely – you may’t simply create a giant database that accommodates one sequence that you simply execute for each doable product, that doesn’t actually work. So the robots must do on-line planning. As a consequence of the truth that we now have a number of robots within the workforce after which additionally the opponents groups’ robots on the sphere, many issues go in another way than deliberate, so a really massive facet of the competitors is execution monitoring and on-line replanning. Principally you create an preliminary plan, however it’s essential continually adapt that plan to what truly occurs throughout execution. I simply need to stress that in distinction to different RoboCup competitions, it’s actually a long-horizon planning process within the sense that we often must do actions on a time horizon of 5 to 10 minutes to truly get to an intermediate purpose of manufacturing a kind of merchandise.

Wataru Uemura: The main focus of our league is on find out how to deal with the manufacturing line. The three cellular robots are an important half. At first these had been autonomous guided robots, however now they’re fully autonomous cellular robots. The robots must resolve on their path to make the product.

May you speak about among the challenges the groups face within the league, and is there a problem or facet of the competitors that the groups have discovered notably troublesome?

Alexander: To start with, one must say that it’s a actually robust downside that we face right here. So the groups which can be beginning new, they need to take care of all the robotic features, so cellular robots, autonomous intelligence techniques, they should drive round, map, work together with the machines. As this stuff that we’re pushing round are non-standard issues, additionally they must construct their very own manipulating units. After which there’s this huge planning facet of the league, which can also be not really easy. After we began this in 2011, 2012, we had been pondering that this have to be a solved downside, that one may use scheduling techniques, that manufacturing is digital, and that all the things could be straightforward. Nonetheless, we discovered it’s not truly straightforward, and there are not any off-the-shelf options for a fleet of robots doing planning and manufacturing.

our workforce (which has turn into notably profitable over time) and from observing the opposite groups as a trustee, I feel that the mixing facet of all of the completely different duties is known as a onerous factor. Having a software program system that’s able to so many issues, speaking with the centralized referee field, and making it run throughout the time restrict of a match, is the main problem. Right now, navigation of a robotic just isn’t the massive challenge, principally, however getting it built-in into your software program system and constructing all the remainder across the planning elements and so forth, that is, from my perspective, the main problem.

Until: As I discussed, I feel the mix of long-horizon planning and execution monitoring is especially troublesome. We do have lots of failures throughout one manufacturing run, due to {hardware} limitations and issues with the robots. Typically the machines themselves fail and they should take care of this, with out having the ability to clear up the issue itself, as a result of it’s not of their management. So they should do lots of reasoning that considers all of the completely different instances that will occur. For instance, instantly you’ve gotten a product showing in a machine and also you now not know the configuration as a result of this info was misplaced on the way in which. How do you take care of this? One other instance is that the robotic drops a bit whereas it’s driving round after which tries to feed it into the machine. Then the machine experiences a failure and the workforce must hold its world mannequin up-to-date to know that this product is now not the place they thought it was. How can we take care of this?

Will there be any new challenges launched for RoboCup 2025?

Alexander: As a result of the issue we are attempting to resolve is so onerous, we don’t have many groups within the league. We had a brand new workforce at RoboCup2023, in Bordeaux, and they’re now persistently coming to the RoboCups, which is good. Other than this, we now have a core of three to 5 groups that take part within the league. Because the problem itself is difficult to get into, the abilities solely develop slowly, so new challenges aren’t actually launched. There are slight adjustments right here and there. One of many main adjustments in recent times was throughout Coronavirus instances the place we needed to abandon the entire match facet as a result of we couldn’t play with two groups on the identical time. We launched features of the sport as challenges. Now we now have a problem monitor as nicely, the place groups can simply deal with sure features of the league and don’t must play the complete sport.

The talents of the groups aren’t creating in such a means that we actually want so as to add new challenges as a result of it’s nonetheless unsolved and onerous for all of the groups to get the robots operating, as a result of we now have so many various features.

I perceive that you’re fascinated with some adjustments to the league. May you say extra about this?

Alexander: Sure, we now have some concepts for a brand new Industrial League. As Festo are pulling out we don’t see the necessity to follow this specific machine sort. We had a workshop earlier this yr with all of the groups and organising committees and we’re proposing a future problem, or league. We are going to talk about this at RoboCup2025.

Until: We truly already had discussions with @work, and agreed that the long-term purpose is to merge the 2 competitions into one massive Industrial League. Subsequent yr, we are going to begin converging by performing some sort of collaboration problem or crossover problem the place groups from @work will collaborate with what’s now the Logistics League, however would be the Good Manufacturing League by that point. Then hopefully, in two or three years, this can be one massive league somewhat than separate industrial competitions. And the concept of the league that we’re at present planning emigrate to is known as a broader good manufacturing situation the place we now have completely different features of good manufacturing. So at present it’s actually solely the manufacturing logistics half. However sooner or later, we additionally need to embody the meeting itself as a part of the competitors and likewise prolong this to humanoid robots and likewise deal with human-robot collaboration on this manufacturing setting.

So this can be very completely different to the Logistics League as it’s proper now. How we are going to do the migration from what we now have to that new league with out dropping all of the groups is one thing that’s nonetheless within the making.

Alexander: I additionally don’t assume we shouldn’t prohibit ourselves to only one sort of robotic. As we see, there’s something occurring with respect to humanoid robots, and the Rescue League is proposing a quadruped robotic. Within the @Dwelling League there are concepts to introduce the staircases the place you want extra agility in these settings. So for me, I wouldn’t prohibit the kind of robots that we’re utilizing. We’re simply fascinated with proposing challenges which can be wanting in the direction of the longer term. To this point we now have been doing issues that we thought is likely to be related to business, however business just isn’t very enthusiastic about what we’re doing right here. Not less than, they aren’t knocking at our doorways and asking what our options are. That’s additionally one other facet that we’ll probably talk about at Robocup 2025 – how we may improve our influence as a league for the surface world.

In order that’s one of many goals, I suppose, to evolve in a means such that business can be extra ?

Alexander: Sure, be related, proper? I imply, you see a lot occurring, particularly in China with the robotic manufacturing OEMs [original equipment manufacturers] that construct these robots in a couple of years which have fairly spectacular capabilities. And nicely, we stand apart and simply watch. So possibly we should always use these robots, and combine them into our course of. It’s essential that we’re opening our minds to check a future that’s completely different from right now.

Concerning the interviewees

	Alexander Ferrein acquired his MSc in Laptop Science (Dipl.-Inform.) and his PhD (Dr. rer. nat) from Aachen College in 2001 and 2007, respectively. Between 2009-2011 he joined the Robotics and Brokers Analysis Lab, College of Cape City, as a postdoctoral analysis fellow with Feodor-Lynen scholarship granted by the Alexander-von-Humboldt Basis. He then re-joined the Data-Primarily based Techniques Group at Aachen College earlier than he grew to become a professor for Robotics and Laptop Science at FH Aachen College of Utilized Sciences. He’s a heading the Cell Autonomous Techniques & Cognitive Robotics Institute at Aachen Utilized Science College. His analysis focusses on the sphere of Synthetic Intelligence and Cognitive Robotics. Since 2015 he’s member of the Advisory Committee of the African-German Community of Excellence in Science whose Vice-president he was between 2019-2023. His analysis concentrates on the sphere of cognitive robotics. Specifically, he’s enthusiastic about high-level management and choice making of robots and brokers appearing below real-time constraints.
	Until Hofmann is a Postdoc at RWTH Aachen College. His analysis focuses on planning, plan execution, generalized planning in addition to reactive synthesis, with a specific deal with planning for robotics. He was a participant within the RoboCup Logistics League from 2016 till 2019 and member of the technical committee from 2017 till 2020. Since 2024, he’s on the manager committee of the RCLL.
	Wataru Uemura was born in 1977, and acquired B.E, M.E. and D.E. levels from Osaka Metropolis College, in 2000, 2002, and 2005. He’s an affiliate professor in Electronics, Info and Communication Engineering Course, School of Superior Science and Expertise, Ryukoku College in Shiga, Japan. He’s a member of IEEE, RoboCup and others. He’s a chairperson of the RoboCup Japanese Nationwide Committee. He’s an govt committee member of RoboCup Logistics League. He was a member of the Industrial Robotics Competitors Committee, the World Robotic Summit. He was TPC Vice Chairs of GCCE 2012, Convention Chair of GCCE 2016, and Publication Chairs of GCCE (International Convention on Client Electronics). He’s a member of the World Abilities in Japan organizing committee of Autonomous Cell Robots.