Dear RoboCupJunior Rescue Maze community,
Congratulations to all on a successful 2023 RoboCup season. We are now starting the development of the 2024 rules and wish to provide you with an overview of planned changes so that we can open up the discussion to the community.
We are satisfied with the current state of the rules, as we feel there aren’t too many elements for new teams to start out, yet field builders are able to create difficult fields that pose a challenge even for experienced teams. It is for these reasons that we decided not to make major changes to the Rescue Maze Field itself, yet there are some exciting planned changes we would like to discuss in the rule drafts.
- We plan to unify the amount of rescue kits dropped for colored and visual victims to Green/U: 0; Yellow/S: 1; Red/H: 2. This should allow for more victims to be placed by the field designer, keeping it under required rescue kits 12, which is the maximum number robots are allowed to carry.
- Obstacles can be placed ANYWHERE in the field, but all the scoring elements are reachable through a passage of at least 20 cm. This means obstacles can be placed in a way that blocks passage to a certain area (or even be taped to a wall), but there is an entrance to said area from a different direction. The robot is still allowed to push obstacles, however it may block its path by doing so.
- The length of the top of the stairs is at least 15 cm (previously 30 cm).
- We also plan to introduce Technical Challenges, which would be a part of the total score (say 10%). These challenges would not necessarily comply with the rules completely, and would require teams to code them at the competition venue. At the same time, we will design these challenges so that no hardware change is required to be made AND the challenge will be simple enough to be coded in a relatively short period of time.
We will be gathering feedback in the following stages:
- August 2023: Major changes released for community feedback
- End of August 2023: 2024 Draft Rules V1 released
- TBD: Open Community Call - Virtual meeting to gather feedback from the community
- September 2023: 2024 Rules Released
Please note that all wording and discussion is not final until the official rule documents are released.
We look forward to receiving your feedback!
2023 Rescue Committee
in general, I think the changes will benefit the Rescue Maze challenge, but I am concerned about the change allowing organizers to place obstacles anywhere in the maze.
First off, if obstacles were allowed to be placed anywhere in the maze, driving around them closely, since there may be only 20cm space, would require not only advanced path finding algorithms, but also a large amount of distance data being extremely hard to provide with the commonly used two-sensor-per-side-layout. Although in theory this is a solvable issue, it brings many new complex hardware and software challenges with it which are partly impossible to overcome for teams using kits limiting the number of sensors (e.g., Lego Mindstorms, Fischertechnik).
Furthermore, allowing obstacles next to walls which, as you explicitly stated, can block certain areas, would be a radical break in the general maze concept. Since the size and shape of obstacles is currently not defined, field builders would be able to design mazes which differ drastically from the current 30x30cm-tile-concept (e.g., mazes with a tile size of 20x20cm, or something in between).
Consequently, navigation algorithms reliant on individual tiles (e.g., DFS, BFS, flood fill, etc.) could not be used anymore to reliably solve the maze. The only alternative I can think of would, be to use some variation of SLAM, but this method requires a decent LIDAR and strong computational capabilities, both of which can be quite expensive and almost impossible to implement for new teams.
I personally believe the current regulation regarding the placement of obstacles in the maze is well balanced. As you pointed out, not overwhelming for new teams, but still allowing field builders to created challenging fields for experienced teams. In my opinion, changing this rule would make it not only be extremely hard to develop a robot able to consistently solve mazes, but more importantly much more difficult for new teams to join and competitively compete in the Rescue Maze competition. At the European Championships, I saw many teams having massive trouble with the current regulations, so making it even more difficult could at worst even prevent teams from competing anymore.
Moritz (Team BioBrause)
Hi Moritz @moritz_biobrause,
Thank You for your feedback regarding this change, I absolutely see the problem this raises, which was not our intention with this change. We definitely wouldn’t want to make 360 LiDARs and complex algorithms like SLAM mandatory for RCJ Maze challenge to be solvable.
Current rules are very limiting in terms of obstacle placement (essentially there have to be four adjacent tiles with no walls at their neighboring edges), thus making obstacles uninteresting and not even challenging for narrower robots, as it can just scoop closer to the wall and keep going. I would like to create a bit more variety with the placement, so instead I propose a different change:
- Obstacles are placed either:
- at least 20 cm from any wall.
- touching any wall and at least 20 cm from the opposite edge of the tile AND any other obstacles. (this means that in this case, obstacle can’t be wider than 10 cm!)
I believe this might create some diversity in both expected procedure upon obstacle detection(move away from the wall than come back, rather than just moving closer to the wall) AND field building, while keeping the challenge simple enough, not requiring any unconventional software nor hardware solutions. Note, that the rule that there cannot be any victims on tiles with obstacles is still valid! Let me know your opinion!
Thank you again for pointing out this issue, it is the exact opposite of what we are trying to achieve!
Hi Matej @Matej,
Thank you for considering my concerns!
In my opinion this regulation looks much better. However, the second case still permits field builders to place obstacles which extend way beyond the wall they are touching. This could potentially cause the same issue described previously since their size and shape is undefined.
Possible solutions might be to either limit the size of obstacles in general or prevent them from extending beyond the tile with the wall they are touching.
Yes, I believe the current iteration does not allow the scenario in your drawing.
“touching any wall and at least 20 cm from the opposite edge of the TILE”
This os the case even if there is no wall in that location. Essentially, obstacle connected to a wall cannot extend away from the wall it is attached to by more than 10 cm.
Therefore, I believe your suggestion with limiting the obstacle size is included in the proposed rule.