Roadmap as seen by TC

[marcomasa]

Educational Competition preparing Students for Industrial Challenges

As we see it, the main focus of RC@Work should be to introduce students to challenges that they might face on their future career path if they decide to stay in robotics. We think that the requirements of our league should encourage students to learn and build, but without setting too high expectations. Therefore we want to restructure some aspects of the competition.

Overall Difficulty

We want to lower the difficulty of the competition to make it easier for new team members to get into a topic. We will try to design the difficulty in a way that allows newer teams to participate in more parts of the competition while also challenging the more experienced students.

Industrial Requirements: Speed & Reliability

We also think that the core concept of a league called "industrial@Work" should focus a bit more on the two key factors of industrial applications: speed and reliability of operation. We therefore want to discard some of the more special cases that are very hard for a robot to perform in general and have therefore not seen much success in the recent years.

Safety of Humans

We continue to allow robots that are not labeled as "safe" by any official institution in order for teams to focus on the research aspect of the tasks. To prevent accidents, we will also continue to set rules that are precautious and will maybe add even more. We will also try to add some ways to enforce such critical rules.

2 Big Steps towards that goal

See #64 and #63

[leanderb]

"should focus a bit more on the two key factors of industrial applications: speed and reliability of operation"

This sentence is somehow confusing. Speed and reliability are already a very important part of a successful competition participation. The mayor collision rules forces the teams to be very reliable in navigation. The object loss rules require the teams to implement reliable object manipulation. Furthermore, the last world cup shows that the duration of a run is rather short, That's why speed is also an existing requirement.

[marcomasa]

Some of the rules were set with that intention, but i think, given the current state of the league, we can go even further in some aspects.

The time bonus has not been rewarded for a long time (maybe except for BMT). To me, this indicates that teams are struggling with the difficulty of the given tasks. This also indicates that right now, we focus more on special / harder tasks, rather than the robot doing some (more simple) tasks reliable and fast.

The past rule changes often made things more difficult. This is mostly due to the fact that the technical challenges have become part of a lot of the competition tests (e.g. arbitrary surfaces, barrier tape etc.). While this was good in the sense of making the tasks more realistic, it also required participating teams to have a lot of knowledge in some research areas.

If we approach the league like a continuous project of a fixed team, the difficulty of the tasks can grow with team experience. But in reality, a lot of RoboCuppers only attend one or two cups during their time in a team. After that, they leave and new (inexperienced) members join, which more often than not cannot fill the gap that the older members left behind.

This is obviously the natural thing that happens, so we should do something to make participation in our league more rewarding for beginners. But even experienced teams often made mistakes during runs, and I think that at least they should be able to do more perfect runs.

As they aren't (by observation) at the moment, I think that the sentence is fitting. It is however quite related to the rest of the roadmap.

[marcomasa]

Maybe this is my personal desire, but I'd like the competition to be decided by

"who had more time boni" (= was faster and more reliable)

rather than (what it is right now)

"who was able to do things at all more often".

At least that was the case when I joined back in 2017 :)

[steup]

I want to add the research aspect to the roadmap, as the generic goal of RoboCup major lies in an integration of education and research, similar to the concept of universities. Therefore, I think we should explicitly state that we want to have competition, that focuses on low entry levels for beginning teams but also offer challenging tasks for experienced teams, which represent the current state-of-the-art or even surpass it.

[steup]

Another point I find very relevant for future extension of the league are new tasks extending the current complexity. In my opinion, there are two generic ways of doing that: The first one involves interaction with new objects, obstacles and environmental components, which may provide new challenges. Examples of this are: the drawer challenge or the assembly challenge. The second type is removal of limitations currently imposed by the rule book to make tasks not too hard. Examples for that are no containers under shelves, and the existence of at least one way to access every workstation at any time. I think both ways increase generality of our approaches and results and is at least as important as reliability if not perfectly controlled environments are considered.