Difficulty of and ideas for LWL #soccer-rules-2023

Hello everybody, @remosgro has posted some intersting experiences relating to the dynamic between LWL and OL with the changes to the ball and the robot size as well as some suggestions over in the main rules thread that I’d like to spin out into it’s own discussion thread here:

Hello Raimondo,

thank you for spending time and though on how to improve LWL. I’ll try to respond to everything block by block and explain my views as well.

@ everyone else: Please feel free to jump in, we’d appreciate any experiences and thoughts you may have.

A higher level of game play is to be expected for the league where the technical challenges are easier to master - IR sensors often run more than 10x the speed of cameras and have higher range and better reliability/robustness, that makes it easier to have high level game play - at least that’s my impression. Following that chain of thoughts the higher level of game play in LWL is to be expected, not a problem and the challenge of OL is to get to the same level with vision as with the “cheat” of the IR ball. In that context the weight limit doesn’t fit the picture too well anymore.

I think this difference is not mainly due to the lack of experience with the new OL rules but mostly due to the light weight robots being able to use their powerful motors because their ball sensing goes 100Hz+. I believe this to be the case because mechanically fast OL robots exist - but they don’t see quickly enough to use it.
Years of development may give some teams an edge but some stability is necessary for schools and clubs to be able to keep up and grow into the rules. There is as always a balance to be struck between teams having access to knowledge and parts at their school instead at the cost of the risk of some re-use of outgoing teams’ work versus needing to buy and learn completely new stuff on a regular basis due to ongoing extensive rule changes. Neither extreme is a great.
The balance of experience, stability and originality will continuously need adjusting. I’m not sure we have enough information to make a definitive call on that yet because it’s the first year RoboCup Junior is back and also because we’re not good at getting feedback in from all over the community, esp. questions with regards to how new teams approach it aren’t easy to evaluate as there may be large differences in different schools, regions and so on.

That is an interesting experience that I would not have expected for the reasons explained above. Does anyone else similar experiences? Please post your experiences below everybody.

I agree that the changes to OL worked well.

We already have the enlarged playing area in the 2023 rules draft increasing the field-size-robot-size-ratio. If it is to be easy for teams to get into we should be careful about reducing how compact they have to make their robots. Such changes (requiring robots to be rebuilt) also require announcements well ahead of time so I suspect it is unlikely to impossible for this to happen for 2023.

We moved away from this ball because it is quite difficult to get similar balls all over the world, Golf balls were how we could solve that problem. The IR ball is now down to one manufacturer of which it is not clear how long it will keep it in production though so that issue might force itself upon us at some point.

I will say about the IR sensing that while it requires to make a board (or rather to have a board made), open source designs made by teams are available and the speed and robustness of recognition lends itself to getting to decent game play much more easily than camera solutions (particularly mirror-based ones that frequently have difficulty recognizing an unobstructed ball at >1m distances.

This has been the topic of some debate. Key points being:

  • voltage limits (the now dead 9V proposal don’t work as higher current motors are available
  • a power measuring device is under consideration but efficiency advantages of expensive motors are significant, esp. over very cheap ones (think 75% vs 15%) so at 20W each the difference would become bigger not smaller
  • Attaching an IMU to approximate how fast robots are going is too sensitive to drift, vibrations, mounting etc.
  • Tracking the robots via a camera over the field is under consideration but it is uncertain all tournament organizers could make this happen
  • we are unsure on how you would implement a tourque limit without building complicated and possibly error-prone measurement devices.

Cheers and thanks again for a very thoughtful contribution,

The 2023 final rules, out now do not contain any concrete changes but we’d like to carry on the discussion for 2024 as LWL does have potential for some more optimization.

Best Regards