Days 67, 70, 71 – Winding Down the Season (03/10/2020)

Authors: Ritik Mishra, Andrew Georgioff

These days were the last days of open room for a while.

Intake Redesign Planning:

At Northern Lights we realized that a wider active intake would be very helpful. One thing that a wider intake would help us achieve is a 5 ball auto that steals the 2 balls that are next to each other from the opponent’s trench.  

Today, we started working on our intake redesign. The current active intake design is 18 inches wide, while our frame is 27 inches, leaving lots of room to become wider. The intake redesign is 23 inches so we can more easily pick up balls, especially those on the rendezvous point. The redesign also makes the corner formed by putting the airpods on the inside of the roller plates is reduced by only using one plate per side instead of two, like the current setup.

Above: An image of the new intake. The roller bar is noticeably wider than the previous design.

Improving shooter accuracy

At Northern Lights, we noticed that our accuracy from the trench was not very high (~50%). 

Since the shooter is very consistent mechanically, a few simple code changes were required in order to boost that accuracy to over 80%. We needed a way to automatically determine the shooter RPM based on how far we are from the target. 

First, we need to know how far we are from the target. Instead of using a fancy equation to determine the distance to the target, we are using an interpolating lookup table that maps from the vertical offset of the target (the ty value given by the limelight) to distance away from the target. The values in this table were measured empirically, and the interpolation seems to be accurate (at least much more accurate than the formula). 

Second, we need to know how fast to spin the flywheel. We also use an interpolating lookup table for this, mapping from distance from the target (provided by the first lookup table) to flywheel RPM. The correct RPM for each distance was also measured empirically by moving the robot to a certain distance from the target, then guessing+checking (by shooting balls) the correct RPM. 

These two lookup tables allow us to automatically adjust the RPM based on our distance away from the target. This also allows us to score balls while moving. As a result, the vision align command has been changed to allow the driver to move closer to/further from the target (as long as they are further than 10 ft away from the target).

Above: A video of autonomous testing from today

Quote of the day: Mecanum heelys… That’s the dumbest thing I’ve ever heard and I’m the one that said it – Andrew Georgioff, Blog Author