FRC 1923: The MidKnight Inventors 2020 Robot - Singularity

FRC 1923, The MidKnight Inventors, are proud to present our 15th year robot: Singularity

Singularity Reveal Video

Robot Highlights:

  • Jam-Free full width intake and indexer
  • 540 degree rotation turret with automated vision tracking
  • High rate of fire shooter powered by 3 Falcon 500s
  • Easy to service upside down electronic panel

Primary Design Constraints:

  • Intake and index balls at maximum possible speed without jams
  • Shoot effectively and quickly from anywhere in the front court without being blocked and from behind the control panel.
  • Be able to intake any of the pre-placed field balls in autonomous, including the ones behind our control panel.
  • Climb next to any partners in every match.

Drivetrain:

  • Eight-wheel west coast drive. 6” wheels. Four omnis, Four AndyMark Performance wheels with Pebbletop Tread.
  • Custom single speed gearbox. Three Falcon 500s per side arranged in a low profile configuration. 14fps Adjusted Speed. Pocketed gears.

Intake:

  • Top and Bottom powered roller intake with carbon fiber rollers to minimize weight.
  • Entire ball path uses polycord to ensure that there are no dead spots.
  • Powered by a custom gearbox with a single Neo550 and an infinity loop of polycord to reverse direction for the bottom roller.
  • Intake geometry optimized to reach 12” outside of frame perimeter to grab balls under the control panel.
  • Intake stowed and deployed using 2 pneumatic cylinders assisted by short stroke gas springs.

Indexer & Ball Tower:

  • Full intake width indexer to serialize balls without jams.
  • Indexer wings powered by a single Neo550 in a custom gearbox which transmits power to both wings such that the wings roll in the same direction (i.e. both to the left, both to the right)
  • Indexer rollers coupled to the gearbox via a timing belt through a one way bearing such that torque is only transmitted to the rollers when the rollers on each respective side are rolling inward (i.e. left roller rolls in while right roller idles and vice versa). This allows us to use one motor and still drive each roller one at a time by oscillating the direction of the motor.
  • Indexer is fast enough to receive 5 balls as fast as the intake can deliver them and still serialize them jam-free
  • Ball tower uses one Neo550 in a custom gearbox to store 5 balls and pass them to the shooter quickly.
  • Ball tower has a combination of polycord and polybelt optimized to reduce jams.
  • Intake, Indexer, Ball Tower are automated for “hands-free” operation as follows:
  • Operator enables indexing function. Rollers start to oscillate. Driver intakes one or more balls up to five. As the intake feeds the balls into the indexer the indexer pushes one ball into the center. A photo reflective sensor detects the ball and stops the oscillation as the center polycord run advances the ball into the robot until a second sensor detects it. This process repeats until commanded to stop manually or the robot detects 5 balls.

Turret & Shooter:

  • Turret is powered by a Neo550 through a VersaPlanetary Lite gearbox. Turret features a custom laser cut 120 tooth delrin sprocket.
  • Turret is geared to rotate 540 degrees in under 1.5 seconds
  • Turret is zeroed with a magnetic limit switch and uses PID control in conjunction with a Limelight 2+ for targeting. PID features compensation for shooting on the move as well as use of the robot’s IMU to help “remember” last known target location while not locked on.
  • Turret service loop is designed to be unplugged at both ends for easy replacement.
  • Shooter is powered by 3 Falcon 500s on a 1.33:1 upduction. Features a 60 Durometer WCP Shooter Wheel and a custom steel inertial wheel.
  • Shooter side plates are made out of delrin for weight savings.

Climber:

  • Climber is inspired by 2056’s 2016 climber. It features a dual 3 stage telescoping tube setup sprung upward by constant force springs held in custom 3D printed spools and retracted by a winch rigged through the inside of each tube.
  • Powered by one Falcon 500 through a custom delrin gearbox. Winches on each tube are linked across the robot by a hex shaft.
  • 3D printed ratchet gear with custom pawl actuated via a rotary pneumatic serves as the brake.
  • Hook assemblies are made out of laser cut delrin.

Control Panel Mechanism:

  • Single 3D printed piece that holds Neo550 through a VersaPlanetary Lite as well as the REV Color Sensor. 3” flex wheel is geared to spin fast enough to spin the control panel at close to the max of 60 RPM.
  • Did not make the final robot due to weight although it could have been added on at a later event.

Technical Binder including Behind the Design, Behind the Controls, and Full Robot CAD coming soon!

31 Likes

Love it. What was the strategic reasoning for going tall instead of short? I’m really curious about this story for a load of teams and find the rationale that everyone used to be interesting. How much of a debate was it and what factors ultimately won out with this design?

1 Like

It was quite a debate for a few days, but one of the biggest factors for going tall was our emphasis on shoot from anywhere in the front court without getting blocked. Since we have the turret to allow us to shoot from any orientation we could not find a good configuration/shooter position for a short bot that couldn’t be blocked when shooting from outside of a safe zone. We also, somewhat correctly, assumed that a lot of teams, especially at high levels of play would build a short bot that would excel at trench run cycles, so we felt that the “clean up” front bot was a valuable role for us to fill with the option to go to the human player station through the rendezvous zone as needed. This also informed our design constraint for a full width intake so we could clean up missed shots and overflow loop balls quickly.

4 Likes

How are you guys dealing with wire management to the turret? 500 degrees is a lot.

1 Like

That fire rate is incredibly satisfying. Reminds me of 469 back in 2006.

9 Likes

We made a fairly stiff turret service cable which was essentially a bundle of all the power and signal wires for the Falcons, and the Ethernet cable for the Limelight zip tied every few inches. We then sheathed that in an abrasion resistant cable sheath. Then we just zip tied the cable to a safe spot on the ball tower and a safe spot on the shooter.

The other trick we used for cable management for the turret is that we wind the cable up around the tower and when we want to shoot, we always spin the turret counterclockwise unwinding the cable until we find the target. Once we find the target we start live tracking it. We help the turret find the target by feeding it information from the robot’s odometry sensors to guesstimate the position of the goal relative to the angle of the turret so we have rough idea of where to look. Then once we fire our volley we return the turret home which winds the cable back up and protects it.

In the event that the cable does get damaged, it has a “universal” connector block on both ends so we can swap out to another pre-made cable.

2 Likes

I weighed in on Twitter about this robot, but to expand on that a little more:

Siiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiick.

2 Likes

Wowwwwww. Eric, Sarath, and team…that robot is incredible. I’m very sad I did not get to see it in person or on the field. Hyped to see the tech binder.

3 Likes

Woah. Can someone confirm that this is the second fastest shooter if not the first? 118 had their 5th ball out before the first went in the goal, 1923 has the capability too

1 Like

Yes, we definitely have the capability for the fifth ball to be released before the first hits the goal. I wouldn’t compare speed with a team that’s already gotten to play just yet, but we’re obviously aiming to be same speed or faster as perennial elite teams like 118.

The “reveal video” footage was filmed during our shooter tuning. Our intent was to devote a day or two of driver practice to get better camera shots of a really dialed in machine, as we weren’t scheduled to compete until week 4. Obviously, we didn’t quite get that far before everything shut down.

When we get to compete, I think it’s fair to make shot for shot comparisons. For now, we definitely have the capability to shoot that quickly - we will see exactly how fast when the robot hits official carpet.

2 Likes

The ability to predict a high level robot and then make the conscious decision to design a robot that compliments that its a very underrated and woefully underused strategy in FRC.

That being said, I wouldn’t count y’all out of the running on being a high seeded robot yourself. The tracking and fire rate your robot has seems to make up for the time spent driving outside of the trench.

Also the name is very apt, considering just how touch-it-own-it your intake is.

1 Like

Thanks! We did definitely target being a high seeded captain at all of our events, and hopefully worlds as well, but the key distinction we made in our thought process for this year was aiming to be complimentary to other elite robots so that we could formulate a stronger alliance. We aimed to be just off meta enough to benefit from what we thought other top teams would converge on.

I felt that this year more so than the last few allowed for more specialization in design without sacrificing the ability to play all aspects of the game at a high level.

4 Likes

I’m assuming that you guys unfortunately didn’t get to play. With that being said, were you guys worried about your omnis breaking if you were to go over the steel barriers in the rendezvous zone?

Yeah, we didn’t quite get that far but we basically had two plans.

  1. Just bring enough omnis to replace every match if needed. Show up to an event with 48 extra omnis.
  2. Custom aluminum omnis made on our CNC router from stacked plates.

That was like a “one week out” problem we knew we wanted to evaluate. Either solution wouldn’t require much effort so we were more focused on tuning our shot and developing autonomous modes the week we shut down. Definitely on the list of things to be prepared for before we competed.

2 Likes

I love the way you put this. It’s something that I have considered and thought about a lot but hadn’t heard it articulated like that.

I think 4414 is comparable

HighTide is one of 1923’s favorite robots of 2020. Loved watching them play because we knew our matchflow would be similar.

What does it take to reach such a fast rate on the balls with a belt system for power cell storage?

From what I understand, you’d want heavy fly wheels not to loose momentum and a high enough speed on the belts. Is your robot using another set of wheels to accelerate the balls going into the turret to provide consistency?

2 Likes

This robot looks great. Definitely in the shooting speed of 118, 4414, 973, 179, etc. Hope to see it compete.

1 Like

High rate of fire and by extension fast cycle time was something we designed around from kickoff day. We started all the way from the intake ensuring that we could plow into a pile of 5 balls on the ground and pull them into the robot without jamming. Then we focused on making sure we could index the balls fast enough to have them ready to shoot in a single file line by the time we arrived at a suitable shooting location. Then we implemented a highly tested ball tower where we iterated on a combination of parameters including polybelt vs. polycord, compression, location and number of flex wheels in relation to belts, and sensor integration. All of this feeds the balls into the shooter through the turret which is perfectly centered over the ball path such that we found additional feeder wheels not needed. Finally the shooter itself has 3 Falcons for more power, and a ~3lb inertia wheel that is pocketed to concentrate the mass around the rim for maximum Moment of Inertia.

3 Likes