Gamification is something that is proposed in many different segments. Bring the entertainment and competition elements to for instance the office environment to collectively drive not only product and service innovation but also organisational changes down to figuring out better models to work with other teams. Or into a social network like HIVE (no words about whether HIVE gamification works or not 😉).
The football aka soccer game is a great gamification approach for Robotics and Artificial Intelligence R&D, I learned recently. The goal: Humanoids that are safe and useful to us humans.
Until yesterday, when I listened to a randomly selected lengthy podcast [1], I didn't realise I had this big ass blind spot. Robo-Football. WOW!
My first encounter with AI was way back, in the early 90s of the last century, when AI was called Fuzzy Logic. Within a team, I worked on a science project for the Air Force; Something to do with the detection of aircrafts in mid-air, by using radar technology. Aiming to lower the false positives; But also to detect stealth planes. Stealth planes are built to not be detectable by radar systems. However one can never be 100% stealthy. A little short bleep on the radar screens is not uncommon. A radar operator (human or machine) would generally classify such a bleep as a false positive. Improvement was needed with more and more stealth fighters hitting the air.
I had no notion about robots and football, no clue at all. Now I am more than curious! Aren't you?
The professor in the podcast is one of those professors I very much like. I don't know him personally. Heck, I hardly know any professor or scientist for that matter, since after my scientific thingies back in the 90s, I went commercial as hell. Nothing to do with science, and more to do with convincing prospects and customers to buy products and services by advising them and designing the (high-level) solutions for them. But I do very much like this professor, simply for the reason that he explained things in pretty normal language. Something important to increase reach, as I learned in my profession. Keep it simple, is the credo.
It seems this professor and his team came to the same realisation after tons of years of robo-football. Keep it simple!
Before they programmed their football players to always know exactly where they were on the football field. They figured it was not important at all to know this, most of the time. With this realisation, they could reduce the number of sensors and amount of computing power to a large degree, freeing up those resources for the more important tasks. Before, to determine the exact position, they used white lines. When they realised the white lines weren't visible in all circumstances (eg another robo-football player was standing on the white line), they increased the number of sensors and the computing power of the robot players. Learning while they go, it was concluded for instance during a battle with a robot player from the opposite team, the exact position is not at all important. The battle needs to be won. Either the robot needs to pass his opponent with the ball still under control, or a pass needs to be executed to another team player. Keep it simple.
Having watched quite a bit of robo-football videos, it sounds very reasonable that football is one of the best, perhaps the best game we know to help the progress of humanoids as well as the progress of rescue robots, industrial robots and home robots. One of the very convincing reasons given is that while the setup of a football game is simple (a small field, defined space, relatively simple rules) the decisions, tasks and actions by the robot players are complex (fast play, team play, opposition).
The idea is to win the game (duh 
One of the newest rules is: At least one pass to another robot player before a score is valid. That rule seemed to be highly complex for AI. From a minimum of 5 goals per match, before this rule was instated, a maximum of 1 goal per match was the result after this became a rule. Team play is one of the key things we need for humanoids. We humans always 'play' in teams. Most of the time we may not even realise. But all day long we deal with other people around us. Action and reaction type of thing. With the added difficulty that our actions and reactions aren't always rational.
Must admit, the podcast left me with more questions than answers. One that hit me immediately regarding location determination: Why not use GPS? With a differential GPS setup. To increase location accuracy to the millimetre levels? I suppose this will not provide the real solution since not everywhere we receive GPS and it is quite cumbersome to redirect such signals into all areas of no line-of-sight with those satellites.
I did like the fact they included methods defined already back in the days I was in University (late 80/early '90s). Such as communication between the robot players. Apparently, with the sensors those players have about 5 meter visibility. So they need the data from the other players to get an overall view of the football field and everything on and in it. Those who researched self-driven cars back in the day included those concepts already (yes, the idea of self-driven cars isn't a recent one, it is already as old as I am, at least 50 years). Commercial planes, those big ones we embark on when we want to move our buds, fly around with systems that let other planes in their vicinity know where they are and who they are. They have done this for ages already, like 40 years, at least. This is for the sole reason to not collide mid-air. When shit hits the fan, two planes head on head, the systems will take over all control and execute a standard manoeuvre to prevent them from hitting each other. I suppose such is needed for humanoids as well since it'll be slapstick all over (good for our laughs and popcorn time) when more than a handful of humanoids mingle with us in the streets.
sidenote: Not being a football fan at all. In the community around another layer 1 chain. I am in this football slash investment slash gamble group. Every weekend we bet on the results of some football league. Leagues from all over the world. When I learned about these robo-football leagues I immediately dropped the idea of including these in our 'project'. That'll be something right? Betting on Robotic/AI games! The leader, being a football buff, someone who bets for the past 14 years on football matches and leagues, was as surprised as I was. He wasn't aware of the robo-players' existence. Those scientist! They can keep their work quiet for the general public!
These robo-football teams came from far. Under the umbrella of RoboCup, they by now not only try to perfectionise the game of football but also develop for other areas such as rescue, help at home and industrial use. To me, this robo-football is to robotics/AI as Formula 1 is to cars. Technology in gameplay driving innovation and optimisation in real-life use cases.
stupidity note: Was wondering when the World Cup is held. Turns out, the 2024 World Cup [6] is played in Eindhoven, a city just 100 kilometres away from where I live. Great! I can see them in real-life action soon! And then the realisation the World Cup is history by now, the tournament held a few weeks ago. Media - again - lagging the real world.
What a wonderful discovery! What do you think? When I know more, I may again order my fingers to hammer my laptop to share newly gained information with you. Unless you say: please don't, your post is a struggle to read let alone to understand 😱 orrrr what about the level of science itself? Sucks: right?
sources
[1] Podcast - the root of all evil 😂 - https://open.spotify.com/episode/10I1IlbPk047jIpWxwdgJn
[2] RoboCup - https://www.robocup.org/
[3] Messi of Football - https://www.tue.nl/en/news-and-events/features/soccer-robots
[4] ARTEMIS the Humanoid - https://www.reuters.com/lifestyle/science/artemis-soccer-playing-humanoid-robot-is-ready-pitch-2023-04-19/
[5] RoboCup Video Library - https://www.youtube.com/channel/UCu0u7DbHJlvATLHjoANa3wQ
[6] RoboCup Worldcup 2024 - https://2024.robocup.org/
all images by edje unless stated otherwise
