At the NC state championship, the angle between the red side wall and the red alliance wall too large by about 10 degrees at the end of finals :ahh: . (Likely the side wall too far out) It got worse over the competition. The position of the hopper was still “correct” based on the dimensions given by FIRST, I guess because the angle is supposed to always be right? I noticed this issue because our robot needed to be set up at a substantial angle off of the wall for it to be parallel to the hopper and hit the hopper in auto just on the red side. (Our auto still worked ) Having to set up the robot without using the wall for alignment significantly lowered the reliability and efficacy of our hopper auto on the red side. The FTAs were excellent and if I had figured this out earlier they would have done their best to fix it. Any ideas for stopping this at champs? I am going to visually examine the field each day and point out any discrepancies to the FTAs, but this angle issue seems very difficult to solve as it might require bending the field.
The field is well designed for what it does… however robots running into it all day WILL cause it to stretch out. And depending on the year, it may stretch a lot. A few years ago, I remember seeing huge bulges in the carpet behind the driver station because of that stretch.
That said, building an autonomous mode that is parametric enough to account for this without multiple iterations of tweaking is a really challenging problem.
Take full advantage of the time allotted for field measurements and calibration. At least in MAR, this usually occurs during day 0, prior to the start of any filler line practice matches. The exact timing may vary by event and competition structure. While we’d all like to have ideally set up fields, this is rarely feasible in reality.
We have a swerve bot and are using motion profiles. Our current auto first drives diagonally so that the robot is 5 in from the wall and the back of the bumpers are 5 in past the hopper plate (where you hit it) then drIves sideways by 20 in at moderate speed. I am planning on switchin this to a bezier curve once some consistancy issues are fixed. (This will have a similar target final position.) Any ideas to improve consistancy?
Very challenging. Teams like ours try to reduce the potential error sources for autonomous modes, but for games like this year with such intricate motions and challenging auto tasks, simply being “astute” often isn’t enough. It takes a good bit of luck and checking to make sure the field is close to what your team expects!
Thanks to all the FTA’s who work to make the fields as close as they can to official spec, match in and match out!
Perhaps a longer list of tolerances should be specified in the game manuals for future years? It would not fix the issues pertaining to inconsistencies. But it would encourage teams to design for them. It would also give the field staff a better idea of when the field is out of spec and requires adjustment. From what I have observed this year, they are kind of being played by ear.
I’d highly recommend this, particularly given the how different field crews assess them at our different events. Also while I agree that higher-tier veterans (folks with hopper autons) adjust for this, having a clear section in the manual with a pretty blue box would help teams that are just looking to make their center gear auto drive far enough. It definitely encourages and reminds teams without as much experience or engineering mentorship that this is part of good design.
I mentioned this in another thread, but we fell victim to this at the Georgian College event. We arrived on Day 0, took field measurements, deployed our code, and went in for a practice match. To our delight (and frankly our surprise) our left-gear auto worked beautifully. The robot swooped in following a cubic spline curve, engaged the peg, dropped off the gear, and drove merrily off toward the center of the field.
The following morning our drive team took another set of field measurements and they were all different. The airships were moved overnight to better match field specifications. Which is fine, and great, but of course as per Murphy’s Law our autonomous mode never really worked well again for the rest of the weekend. (It in fact turned out not to be a problem with the field dimensions, but a faulty encoder that was giving us wonky, and therefore unpredictable, results.)
Still, the exasperated cry from our software team is perhaps the most memorable part of the weekend: “Our autonomous worked perfectly, and then they moved the field!”
I’m a little confused as to what the pictures are showing. I see that the tape lines are hitting the black plate at different spots, but that doesn’t give a great indication of what’s actually wrong - the lines could be measured wrong, plate could have been placed wrong, carpet could have shifted, field border could have shifted, field could have been built slightly out of square…or all of the above. If you hadn’t mentioned the angle between the border and the wall being off, I would have just assumed that someone measured wrong when doing the tape lines.
Unfortunately there are some things that the field staff can’t fix once the field is together - such as the field border (unless part of it flat-out breaks). The setup crew should be making every effort to set up the field walls and border straight and square, because once it’s together, it’s together, and it can’t be significantly adjusted (by humans, in a proper way that avoids damaging the field border) unless something is really off. If the FTAs and the rest of the field staff can do anything to fix things like the border shifting or carpet being stretched, they will, but it’s usually out of their control.
I think this is a great idea, as long as it comes with a disclaimer that some aspects of the field simply cannot be fixed mid-competition, such as carpet bubbles. I don’t think there’s an accurate “safe” tolerance that will cover 95% of inconsistencies with the dimensions/locations of field components (well, there is, but it’d be like…+/-1.5"), which is probably why FIRST doesn’t provide a better list. The best thing that teams can do at the moment is take advantage of the field measurement period and let the FTAs know if they find anything they feel is out of spec. If there’s something that’s off, it’s more likely that it can be fixed before matches start than after. Maybe it’s different in other districts/regionals, but in my experience with FiM and MAR, surprisingly few teams take advantage of this time period.
I actually did this year, and personally taped diagonal lines like the one shown in the OP.
Maybe I was missing something, but the dimensions provided for a ton of stuff, like taping and Hopper locations, were pretty lacking. The diagonal lines for the Key and Retrieval Zone were both based on distances from the face of the Boiler or loading stations. It’s a lot easier to draw a straight line in CAD than it is to mark points in a sea of carpet, then perfectly extend out from them. I guess that for better field setup the volunteers could try to whip out some CAD and find more dimensions or use some surveying equipment to try to extrapolate from a couple points marked relative to a different field element.
An easier and more consistent solution to some of the problems would be providing more detailed field drawings with more dimensions. I’m sure the pictures posted in this thread are from a lot of different things, but at least some of them would be solved with better drawings. Given how much of this field has a tendency to break, I would think that you would want to make something like taping lines and positioning Hoppers really easy for field setup and field reset crews, who have to quickly replace torn up tape and move field elements that shifted on their velcro. It would also help to not have so many different field elements that can break or somehow get out of spec, but it’s a bit late for that.
Do you know this is the case? I’d be surprised to hear that an event moved their airships after they were set. Those are not easy things to move! Obviously out of spec is out of spec and the field staff would need to address that.
To those with code that relies on particular placement of elements I always recommend additional sensing such as vision tracking. Options like spline curves are great for getting close, and vision tracking can help ensure that once you are close you hit dead on.
As an added bonus, this is more “real.” You can take all the pictures and measurements of Mars you want before you launch your rover, but by the time it gets there the wind will have moved everything around a bit. The universe rarely stays constant.
I do wish the tolerances on some of the important (or likely to move) elements would be a bit more clear- the field diagrams are (I believe) the only place the tolerances are specified, and those aren’t always the easiest things to use for finding such information.
I feel your pain. The amount of extrapolating required to tape out my team’s field was scary at times, especially when the dimensions provided by FIRST and the CAD dimensions conflicted. Just a few more “official” dimensions would have saved several hours of frustration.
We had to move both the airships a little more than a foot toward the center of the field at one of the events. That’s what happens when you listen to someone else calling numbers and don’t check them.
In the morning we measured the right line and moved the airships, the whole process only took about 30 minutes for both? It’s not something I want to do again, but it wasn’t nearly as bad as it looks like it would be.
At risk of dragging this thread down a tangent, the timing of this period may be part of its lack of use. In MAR at least, this is usually during the day 0 of the event during the early evening. As a result, sometimes teams will not have arrived at the venue until the field measurement period has started (or even concluded), and their initial time is often dedicated to loading in, checking in, getting their robot unbagged, setting up their pit, setting up their radio, and getting inspected. I know this is often the case for 1712, as we have to wait until after our public school busses complete their normal routes before we can use one to transport our team/robot to the event (which in turn often puts us at the mercy of rush hour traffic). There are a few other teams that completely skip day 0, opting to show up as pits open on day 1.
I was at Hartford this past weekend, and I noticed when I got there on Saturday that the line for Key overlapped the Hopper trigger on one side of the field by several inches, and not overlapped at all on the other side. I told the Field Supervisor, and he double-checked all the measurements, and did end up moving one of the lines slightly, but he claimed it met all the dimensions given in the field setup documentation. I have no reason to doubt him.
There’s a paragraph in the manual that’s been there as long as I can remember (my emphasis added):
Could things be better and easier specified? Sure. Is it annoying when the field changes size and shape throughout an event due to robots banging on all the components? Yup. Volunteers do what we can, and in general are always welcoming to practical suggestions to help keep the field in shape. But in general, consider it a lesson in real-life engineering, and work to adapt as best you can.
) Having to set up the robot without using the wall for alignment significantly lowered the reliability and efficacy of our hopper auto on the red side. The FTAs were excellent and if I had figured this out earlier they would have done their best to fix it. Any ideas for stopping this at champs? I am going to visually examine the field each day and point out any discrepancies to the FTAs, but this angle issue seems very difficult to solve as it might require bending the field.