Presentation of the perfect paper award on the RoboCup 2025 symposium.
An essential facet of autonomous soccer-playing robots issues correct detection of the ball. That is the main target of labor by Can Lin, Daniele Affinita, Marco Zimmatore, Daniele Nardi, Domenico Bloisi, and Vincenzo Suriani, which gained the perfect paper award on the latest RoboCup symposium. The symposium takes place alongside the annual RoboCup competitors, which this 12 months was held in Salvador, Brazil. We caught up with a number of the authors to search out out extra in regards to the work, how their methodology could be transferred to functions past RoboCup, and their future plans for the competitors.
May you begin by giving us a quick description of the issue that you simply had been making an attempt to resolve in your paper “Self-supervised Function Extraction for Enhanced Ball Detection on Soccer Robots”?
Daniele Affinita: The primary problem we confronted was that deep studying typically requires a considerable amount of labeled knowledge. This isn’t a significant downside for frequent duties which have already been studied, as a result of you may normally discover labeled datasets on-line. However when the duty is extremely particular, like in RoboCup, you should gather and label the information your self. Meaning gathering the information and manually annotating it earlier than you may even begin making use of deep studying. This course of just isn’t scalable and calls for a big human effort.
The thought behind our paper was to scale back this human effort. We approached the issue by means of self-supervised studying, which goals to study helpful representations of the information. In spite of everything, deep studying is basically about studying latent representations from the obtainable knowledge.
May you inform us a bit extra about your self-supervised studying framework and the way you went about creating it?
Daniele: Initially, let me introduce what self-supervised studying is. It’s a manner of studying the construction of the information with out getting access to labels. That is normally executed by means of what we name pretext duties. These are duties that don’t require specific labels, however as an alternative exploit the construction of the information. For instance, in our case we labored with photos. You may randomly masks some patches and prepare the mannequin to foretell the lacking elements. By doing so, the mannequin is pressured to study significant options from the information.
In our paper, we enriched the information through the use of not solely uncooked photos but in addition exterior steerage. This got here from a bigger mannequin which we consult with because the trainer. This mannequin was skilled on a unique activity which is extra common than the goal activity we aimed for. This manner the bigger mannequin can present steerage (an exterior sign) that helps the self-supervision to focus extra on the precise activity we care about.
In our case, we needed to foretell a good circle across the ball. To information this, we used an exterior pretrained mannequin (YOLO) for object detection, which as an alternative predicts a unfastened bounding field across the ball. We will arguably say that the bounding field, a rectangle, is extra common than a circle. So on this sense, we had been making an attempt to make use of exterior steerage that doesn’t resolve precisely the underlying activity.
Overview of the information preparation pipeline.
Have been you capable of check this mannequin out at RoboCup 2025?
Daniele: Sure, we deployed it at RoboCup 2025 and confirmed nice enhancements over our earlier benchmark, which was the mannequin we utilized in 2024. Particularly, we seen that the ultimate coaching requires a lot much less knowledge. The mannequin was additionally extra strong below totally different lighting situations. The problem we had with earlier fashions was that they had been tailor-made for particular conditions. However after all, all of the venues are totally different, the lighting and the brightness are totally different, there could be shadows on the sphere. So it’s actually essential to have a dependable mannequin and we actually seen an ideal enchancment this 12 months.
What’s your staff identify, and will you speak a bit in regards to the competitors and the way it went?
Daniele: So our staff is SPQR. We’re from Rome, and we now have been competing in RoboCup for a very long time.
Domenico Blois: We began in 1998, so we’re one of many oldest groups in RoboCup.
Daniele: Yeah, I wasn’t even born then! Our staff began with the four-legged robots. After which the league shifted extra in the direction of biped robots as a result of they’re tougher, they require steadiness and, total it’s tougher to stroll on simply two legs.
Our staff has grown rather a lot throughout latest years. We now have been following a really optimistic development, going from ninth place in 2019 to 3rd place on the German Open in 2025, and we bought 4th place at RoboCup 2025. Our latest success has attracted extra college students to the staff. So it’s sort of a loop – you win extra, you appeal to extra college students, and you may work extra on the challenges proposed by RoboCup.
SPQR staff.
Domenico: I need to add that additionally, from a analysis viewpoint, we now have gained three greatest paper awards within the final 5 years, and we now have been proposing some new traits in the direction of, for instance, using LLMs for coding (as a robotic’s behaviour generator below the supervision of a human coach). So we try to maintain the open analysis subject energetic in our staff. We need to win the matches however we additionally need to resolve the analysis issues which are sure along with the competitors.
One of many essential contributions of our paper is in the direction of using our algorithms exterior RoboCup. For instance, we try to use the ball detector in precision farming. We need to use the identical method to detect rounded fruits. That is one thing that’s actually essential for us; to exit the context of Robocup and to make use of Robocup instruments for brand spanking new approaches in different fields. So if we lose a match, it’s not an enormous deal for us. We wish our college students, our staff members, to be open minded in the direction of using RoboCup as a place to begin for understanding teamwork and for understanding find out how to take care of strict deadlines. That is one thing that RoboCup may give us. We attempt to have a staff that’s prepared for each sort of problem, not solely inside RoboCup, but in addition different varieties of AI functions. Profitable just isn’t all the pieces for us. We’d choose to make use of our personal code and never win, than win utilizing code developed by others. This isn’t optimum for reaching first place, however we need to train our college students to be ready for the analysis that’s exterior of RoboCup.
You mentioned that you simply’ve beforehand gained two different greatest paper awards. What did these papers cowl?
Domenico: So the final two greatest papers had been sort of visionary papers. In a single paper, we needed to provide an perception in find out how to use the spectators to assist the robots rating. For instance, in the event you cheer louder, the robots are inclined to kick the ball. So that is one thing that’s not really used within the competitors now, however is one thing extra in the direction of the 2050 problem. So we need to think about how it will likely be 10 years from now.
The different paper was referred to as “play in every single place”, so you may, for instance, play with several types of ball, you may play exterior, you may even play with out a particular aim, you may play utilizing Coca-Cola cans as goalposts. So the robotic has to have a common method that’s not associated to the precise subject utilized in RoboCup. That is in distinction to different groups which are very particular. We now have a unique method and that is one thing that makes it tougher for us to win the competitors. Nonetheless, we don’t need to win the competitors, we need to obtain this aim of getting, in 2050, this match between the RoboCup winners and the FIFA World Cup winners.
I’m taken with what you mentioned about transferring the strategy for ball detection to farming and different functions. May you say extra about that analysis?
Vincenzo Suriani: Our lab has been concerned in some totally different initiatives referring to farming functions. The Flourish challenge ran from 2015 – 2018. Extra lately, the CANOPIES challenge has focussed on precision agriculture for everlasting crops the place farmworkers can effectively work along with groups of robots to carry out agronomic interventions, like harvesting or pruning.
We now have one other challenge that’s about detecting and harvesting grapes. There’s a large effort in bringing information again from RoboCup to different initiatives, and vice versa.
Domenico: Our imaginative and prescient now’s to deal with the brand new technology of humanoid robots. We participated in a brand new occasion, the World Humanoid Robotic Video games, held in Beijing in August 2025, as a result of we need to use the platform of RoboCup for different kinds of functions. The thought is to have a single platform with software program that’s derived from RoboCup code that can be utilized for different functions. If in case you have a humanoid robotic that should transfer, you may reuse the identical code from RoboCup as a result of you should utilize the identical stabilization, the identical imaginative and prescient core, the identical framework (roughly), and you may simply change some modules and you may have a totally totally different sort of utility with the identical robotic with roughly the identical code. We need to go in the direction of this concept of reusing code and having RoboCup as a check mattress. It’s a very robust check mattress, however you should utilize the ends in different fields and in different functions.
Wanting particularly at RoboCup, what are your future plans for the staff? There are some large adjustments deliberate for the RoboCup Leagues, so may you additionally say how this may have an effect on your plans?
Domenico: We now have a really robust staff and a number of the staff members will do a PhD within the coming years. One among our targets was to maintain the scholars contained in the college and the analysis ward, and we had been profitable on this, as a result of now they’re very passionate in regards to the RoboCup competitors and about AI usually.
When it comes to the adjustments, there can be a brand new league inside RoboCup that could be a merger of the usual platform league (SPL) and the humanoid kid-size league. The humanoid adult-size league will stay, so we have to determine whether or not to hitch the brand new merged league, or transfer to adult-sized robots. In the intervening time we don’t have too many particulars, however what we all know is that we’ll go in the direction of a brand new period of robots. We acquired robots from Booster and we are actually buying one other G1 robotic from Unitree. So we try to have an entire household of latest robots. After which I believe we are going to go in the direction of the league that’s chosen by the opposite groups within the SPL league. However for now we try to prepare an occasion in October in Rome with two different groups to trade concepts and to grasp the place we need to go. There will even be a workshop to debate the analysis aspect.
Vincenzo: We’re additionally in dialogue about the perfect dimension of robotic for the competitors. We’re going to have two totally different positions, as a result of robots have gotten cheaper and there are groups which are pushing to maneuver extra shortly to a much bigger platform. Then again, there are groups that need to persist with a smaller platform so as to do analysis on multi brokers. We now have seen loads of functions for a single robotic however not many functions with a set of robots which are cooperating. And this has been traditionally one of many core elements of analysis we did in RoboCup, and in addition exterior of RoboCup.
There are many factors of view on which robotic dimension to make use of, as a result of there are a number of components, and we don’t understand how quick the world will change in two or three years. We try to form the principles and the situations to play for subsequent 12 months, however, due to how shortly issues are altering, we don’t know what the perfect resolution can be. And likewise the analysis we’re going to do can be affected by the choice we make on this.
There can be some adjustments to different leagues within the close to future too; the small and center sizes will shut in two years in all probability, and the simulation league additionally. Quite a bit will occur within the subsequent 5 years, in all probability greater than over the past 10-15 years. This can be a important 12 months as a result of the choices are based mostly on what we will see, what we will spot sooner or later, however we don’t have all the knowledge we want, so it will likely be difficult.
For instance, the SPL has an enormous, in all probability the largest, group among the many RoboCup leagues. We now have loads of groups which are grouping by curiosity and so there are groups which are sticking to engaged on this particular downside with a selected platform and groups which are making an attempt to maneuver to a different platform and one other downside. So even inside the identical group we’re going to have a couple of viewpoint and hopes for the longer term. At a sure level we are going to strive to determine what’s the greatest for all of them.
Daniele: I simply need to add that so as to obtain the 2050 problem, in my view, it’s essential to have only one league encompassing all the pieces. So up up to now, totally different leagues have been specializing in totally different analysis issues. There have been leagues focusing solely on technique, others focusing solely on the {hardware}, our league focusing primarily on the coordination and dynamic dealing with of the gameplay. However on the finish of the day, so as to compete with people, there should be just one league bringing all these single elements collectively. From my viewpoint, it completely is smart to maintain merging leagues collectively.
Concerning the authors
![]() |
Daniele Affinita is a PhD pupil in Machine Studying at EPFL, specializing within the intersection of Machine Studying and Robotics. He has over 4 years of expertise competing in RoboCup with the SPQR staff. In 2024, he labored at Sony on area adaptation methods. He holds a Bachelor’s diploma in Pc Engineering and a Grasp’s diploma in Synthetic Intelligence and Robotics from Sapienza College of Rome. |
![]() |
Vincenzo Suriani earned his Ph.D. in Pc Engineering in 2024 from Sapienza College of Rome, with a specialization in synthetic intelligence, robotic imaginative and prescient, and multi-agent coordination. Since 2016, he has served as Software program Improvement Chief of the Sapienza Soccer Robotic Staff, contributing to main robotic competitions and worldwide initiatives corresponding to EUROBENCH, SciRoc, and Tech4YOU. He’s at present a Analysis Fellow on the College of Basilicata, the place he focuses on creating clever environments for software program testing automation. His analysis, acknowledged with award-winning papers on the RoboCup Worldwide Symposium (2021, 2023, 2025), facilities on robotic semantic mapping, object recognition, and human–robotic interplay. |
![]() |
Domenico Daniele Bloisi is an affiliate professor of Synthetic Intelligence on the Worldwide College of Rome UNINT. Beforehand, he was affiliate professor on the College of Basilicata, assistant professor on the College of Verona, and assistant professor at Sapienza College of Rome. He obtained his PhD, grasp’s and bachelor’s levels in Pc Engineering from Sapienza College of Rome in 2010, 2006 and 2004, respectively. He’s the writer of greater than 80 peer-reviewed papers printed in worldwide journals and conferences within the subject of synthetic intelligence and robotics, with a deal with picture evaluation, multi-robot coordination, visible notion and data fusion. Dr. Bloisi conducts analysis within the subject of melanoma and oral carcinoma prevention by means of automated medical picture evaluation in collaboration with specialised medical groups in Italy. As well as, Dr. Bloisi is WP3 chief of the EU H2020 SOLARIS challenge, unit chief for the PRIN PNRR RETINA challenge, unit chief for the PRIN 2022 AIDA challenge. Since 2015, he’s the staff supervisor of the SPQR robotic soccer staff taking part within the RoboCup world competitions |
![]() |
Can Lin is a grasp pupil in Information Science at Sapienza college of Rome. He holds a bachelor diploma in Pc science and Synthetic intelligence from the identical college. He joined the SPQR staff in September of 2024, specializing in duties associated to pc imaginative and prescient. |
Lucy Smith
is Managing Editor for AIhub.