lol - From a shitless slave region of humanism being suffered under hoon-tribe-originated-parasitism and kiki'she' collectivism which is known as the name 'Korea', whose dwellers always steals and begs from Japan's original history which clearly inherited sincere humanity in Ancient Cudara more than its 'original' dwellers in hometown.
I've been following this cup for 15 years... and honestly, improvements have been very minor during this period. I know it's a hard and challenging field (I'm a robotic engineer), but the developments in machine learning, computer vision and processors should have had a more obvious impact. The goal of defeating the human team in 2050 seemed realistic 15 years ago ; now it's just utopic.
Just as a comparison, humanoid robots at Robocup 2003. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-MVcDOSsdlu4.htmlm39s I'd say there might have been a little bit of progress ;)
What you share wasn't really the final match (with, normally, the best teams).... 15 years ago might be a little mean, but if you look at the finals 10 years ago, the difference is very little (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-iMM_XQXJUUc.html (2008) , ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-xjSm4nE5lcY.html (2009) ) I would love to see a match between the algorithms 10 years ago and the ones today ; and I would love if they can prove I'm wrong
A few things that changed between the 2009 game and the 2018 one: - blue and yellow goals got replaced by white ones (much harder to detect, also you can't tell by the goal color anymore which of the two goals you have to score in) - orange ball was replaced by black and white ball (also much harder to detect since all other things on the field are black, white, or grey) - field size got increased from 6x4 to 9x6 (makes localizing much harder and makes it more important to play as a team) - 3 vs 3 became 5 vs 5 (increased need for team play to make sure there's exactly one robot handling the ball and the others don't run into each other) - jerseys are completely custom, back then they were blue and red and made it much easier to detect and differentiate opponents and your own team - the floor is now an 8mm artificial grass instead of a billiard table-like carpet which is a lot harder for walking - robots have to now listen to the referee's whistle and don't get told over wifi when the game starts (and they have to make sure they don't react to the whistle on a neighbor field when there's a game running there, too) Back in 2008 it was easy to get good results by just having the simplest strategy possible: turn around until you see a ball, if you do walk to it, when you're there turn around the ball until you see the blue or yellow goal, when you do shoot. By now you won't get far using that, you'll need a good team strategy. There have also been a lot of improvements in terms of ball handling. You can see that the robots now are much quicker when they're close to the ball. Also, considering that the servos and joints have stayed nearly the same we're moving much faster now on a much more difficult turf. So in short: the algorithms from 10 years ago wouldn't be able to walk on the new field, they wouldn't know where they are because they lines are too far away, wouldn't know where the ball is and even if they hit it would have no idea which direction to shoot to.
Yes, we're improving speed all the time. You can check the World Championship 2018 final video ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-H8xc6LpiNVs.html where we improved reaction time at the ball quite a lot. For next year we'll have a new version of the robot with a faster processor that will hopefully increase speed further (though the servos are about the same, so overheating will still be a very limiting factor).
We (the blue team) were called for "pushing" the goalie, however the ball was between the robots, so it shouldn't have been a penalty... But our No 2 was close by and finished the job anyway :D
All the teams are equal in their hardware (as in the robots are just bought from a company). However each team develops their own software. The robots need to act autonomously, there is no remote control. So each team needs to write software to make them walk (and not fall down), see the ball with the robot's cameras, see the field, the lines on the field, the other players, get a sense of direction so they know where on the field they are. Then they have to communicate (via Wifi) with their robot teammates and then try to find a good strategy as a team to win. And all of this has to work together. If a single thing fails your robots can't play. Like if they can't detect the ball they can be the fastest robots but won't score a goal. Or if they are fast and can detect the ball but don't know where on the field they are they might even score into their own goal. ;)
Was comparing this with robot football from 10 years ago - RoboCup 2008 given below: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE--CKaJHd_Eis.html Hardly any difference that I could see. Based on this and similar evidence, it looks like despite huge efforts, the advancement in robotics has been practically insignificant in the last 10 years, at least in the area of human-like locomotion.
The second one. It also helps with keeping the hip joints save when falling backwards since they'll land on their fingers (and break those) instead of their hip. ;) We like to think of it as Naruto style. ;) ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-pPeACdLYbDk.html
