Today is the day, the team is loading in the robot into the Waterloo event later in the day and although sorry for the lack of updates the past 2 weeks with a mid-semester break and a busy couple of days preparing afterwards, we have made some strides into improving our robot for Waterloo, and we are really excited for what’s to come in this event!
Upon the conclusion of the Georgian event, we set a couple of goals we want to achieve going into waterloo, and these included:
Improved and Faster Auto Routines
We have successfully implemented vision and pose estimation to the system after a moderately painful week of debugging, tuning and planning. As a result we have devised 3 main autos to work with for this event; an L2 2.5 Game piece + Balance, an L2 3 Game piece “Link” Auto, and a cable protector side auto.
More Human Player - Driver/Operator Coordination and Better HP Cube Pickup
Had a series of practice runs to help familiarize HP/Driver coordination more, leading to less cone handoff issues and a faster cube HP Intake run.
Shelve the Telescoping Arm, Keep and Improve the current System
Kept our goal which was pretty easy and simple, although the CNC’d telescope extrusions were delivered to us (it looks so nice oh my, see below for a photo of it), we plan on using it as an offseason project instead…
Faster Cycling and More Consistent L3 Cone Scoring
Unfortunately a more consistent L3 cone scoring was found to be difficult and requires large enough mechanism changes that would eat too much time and undo some progress software has made(given we only had about a week to work with since the other week was march break for us), so we decided to only do L3 cones on an extraordinary basis for this event.
However we were able to do a series of driver practice runs that improved our cycle times at endgame relative to what we were already achieving at Georgian, an important thing to note though is that this was done with no defense, so we will see how much defense will bring down that number for us.
Cone and Cube Intake position improvements
We have noticed that our cube and upright cone intake setpoints at georgian were suboptimal, leading the intake to miss slightly overinflated cubes or tip upright cones, rather than intake them in. So we introduced some minor changes to improve their reliability.
We also tried installing centering guides into the cone intake, although albeit a bit fragile in the early iterations, it works well in centering the cone, avoiding the variations associated with the roller intakes and resulting in a faster and more consistent teleop and auto cone scoring.
Below are a series of cycling tests that we have done, again huge thanks to 5406 for letting us use their practice field.
These tests allowed us to find out that we were able to cycle a full L2 grid + some L1 or L3 cube in one full match (assuming a 2 game piece auto), which is really promising but as noted above, this is without defense so we will see.
We also found out that in our case, the fastest way to fill L1 is using cones and not cubes.
One of these videos had another person manning HP instead of our HP, and it shows that practice and coordination is really important if we want a fast and consistent human player cycle time.
In some of the videos, it seemed that we had difficulty intaking ground tipped cones, which is not that much of a deal given our focus is more on chute pickup at the moment, but it was later found to be a wrong setpoint.
We have found out that the inconsistency associated with our L3 Cone is too high, which is why we do not really want to use it unless really needed. Something to look onto again and solve after Waterloo though.
We initially had issues about having to mirror pathplanner pathing, causing issues and possibly requiring us to do 2 separate paths per side for an auto, this issue took a while to fix but was eventually fixed by creating a custom python script that just mirrors the path.
Videos of some of the autos we have done, playing with various times and techniques so we can obtain a faster and more consistent autonomous
L2 “Link” run, even though it lasts slightly more than 15s in this video, we managed to get it down to about 13s at the conclusion of yesterday.
L2 2 game piece balance runs, a series of videos showing numerous attempts at trying to test this auto mode
One thing to note is that this practice field is in rough shape. One of our mentors had to hold the charge station up so it didn’t get stuck when we climbed (some of the videos they failed). The barrier isn’t straight and moves so it’s not even easy to align correctly at the start and there is only 1 mark to place game pieces down (where the cube was) so the cone was measured and best guessed.
So a lot of those near misses are seeming like they are poor field alignment. So hopefully later after load in today, we can get practice matches in to make little tweaks to the pathing.
Overall we are very happy with the results, this is the most advanced auto setup the team has had in its 13+ year history and it is something that is very significant to the team.
Other Stuff that happened in the 2 weeks
We have made pit collateral hopefully done in time for this event, showing the basic features of the robot when explaining them to judges or interested folks in the pits.
Also our 2023 robot page is up at: 2023 Season | Team 3161 - Tronic Titans
Telescope Extrusions CNC’d by a sponsor, unfortunately these won’t be on the bot for the official season, hopefully in the offseason it will be!
We swapped our green compliant wheels for custom gold wheels for the rest of the season.
Found a potential failure mode in our intake belts, we believe this is due to a lack of current limiting in the system and have implemented changes to prevent that for Waterloo.
This week was full of work and dedication in order to prepare for our next event, and was very neat seeing what other teams have done for week 2 and week 3, we are really excited for what Waterloo has to offer since there are a lot of very strong teams in the lineup. As always, lots of luck and love to all of the competing teams this year. Make sure to check out our build thread soon for any other robot updates and our waterloo recap post!