Sally & Cruz · Push Back Gamma Robots

Luigi and Guido could still perform well, but continued iteration could not remove the fundamental restrictions created by their architecture.

The major limitations included:

• Tank drive restricted lateral movement and field positioning
• Limited acceleration made the robots easier to defend and pin
• The robots could not drive underneath the Long Goal
• Double parking was difficult and sensitive to the approach
• Block compression caused recurring indexer and outtake jams
• Wings were added after the original architecture was complete
• Several systems were optimized around Robot Skills rather than Worlds match play

The goal of Sally and Cruz was to create a pair of robots with a substantially higher competitive ceiling rather than only correcting individual reliability problems.

The Gamma robots needed to:

• Move holonomically
• Accelerate faster during match play
• Drive beneath the Long Goal
• Use a wider, lower-compression Block path
• Carry at least 7 Blocks, with a target capacity of 10
• Sort Blocks by color
• Score into all required Goals
• Complete faster autonomous scoring cycles
• Use a robust wing designed into the original architecture
• Double park more reliably
• Drive through the Park Zone to collect Blocks
• Complete the Autonomous Win Point
• Retain the team's custom localization architecture
• Use the same core design for the 24" and 15" robots
• Remain under approximately 20 lbs, with a preferred target below 15 lbs

Sally and Cruz were designed around Worlds match play rather than early-season Skills qualification.

The primary strategic changes were:

• Holonomic movement
• Driving beneath the Long Goal
• Faster drivetrain acceleration
• Front-to-back Block flow
• A wider Block path with less compression
• Color sorting
• Integrated wings
• Improved double parking
• Faster autonomous cycles
• Greater control around the Center Goals

The opposite-side intake and outtake layout allowed the robots to collect Blocks from a Match Loader and drive directly toward a Long Goal without turning around.

Holonomic movement allowed the robots to make lateral adjustments while maintaining their orientation. This supported Goal alignment, defensive movement, under-Goal driving, and the use of a single wing instead of requiring matching mechanisms on both sides.

Sally and Cruz shared the same core subsystem architecture. The frame dimensions changed between the 24" and 15" versions, but the drivetrain concept, Block path, scoring systems, wings, electronics, and autonomous strategy remained the same.

Each robot used 12 drivetrain motors:

• 10 motors for forward and backward movement
• 2 motors for lateral strafing

The 10-motor forward/back drivetrain used a slightly lower wheel speed than Luigi and Guido, but the additional motors produced faster acceleration. This was more valuable during match play than a higher theoretical top speed because the robots could change direction and reach useful speeds sooner.

The strafe axis was initially designed to be speed-matched to the approximately 480 rpm forward/back drivetrain. Testing showed that the rocker pod did not have enough torque at that ratio, so the strafe axis was reduced to approximately 160 rpm.

The H-drive allowed the robots to:

• Move laterally without rotating
• Align with Goals and Match Loaders
• Reposition beneath the Long Goal
• Escape defensive pressure
• Adjust wing position through strafing
• Support more complex autonomous paths

The drivetrain packaged its motors around short 0.875" shafts. Thin outer bearing plates supported the shafts while also serving as finished exterior plates.

The strafe wheels were mounted in a rocker pod that remained above the field while inactive and deployed when the strafe motors applied torque.

The first version did not deploy consistently. Its gear arrangement produced an ineffective reaction force, and manually pushing the wheels into the floor caused them to damage the tiles instead of moving the robot.

I added another idler stage to reverse the wheel direction relative to the motors. This changed the torque reaction acting on the rocker and caused the pod to rotate downward into the field.

The intended 480 rpm ratio still did not provide enough torque, so the strafe output was reduced to approximately 160 rpm. The slower output improved deployment force and made lateral movement possible, although the system still needed additional tuning.

The robots needed to remain short enough to drive underneath the Long Goal while packaging 12 drivetrain motors, a two-wide Block path, pneumatic mechanisms, electronics, and odometry hardware.

Weight-saving decisions included:

• Pocketed aluminum plates
• Thin sheet material where appropriate
• Polycarbonate structural and mechanism components
• 3D-printed mounts, funnels, spacers, and housings
• Shared structural plates between subsystems
• Compact motor and electronics packaging
• Short drivetrain shafts
• Custom bearing supports

The low frame opened movement and strategy options that were unavailable to Luigi and Guido, but it also made subsystem packaging and service access more difficult.

The intake collected Blocks from the floor and fed them into the two-wide indexer.

Its geometry supported:

• Match Loader collection
• Park Zone Block collection
• Center rush strategies
• Front-to-back Block flow
• One-time deployment from the starting configuration

Placing the intake opposite the outtake allowed the robots to collect Match Loaded Blocks and drive directly toward a Long Goal without turning around.

The intake was also used for lower Center Goal scoring. Reversing it pushed Blocks back out through the front of the robot and into the Goal.

The two-wide indexer was designed to reduce Block compression while maintaining useful capacity and throughput.

Its goals included:

• Carrying at least 7 Blocks
• Targeting a capacity of up to 10 Blocks
• Reducing jamming
• Moving multiple Blocks through a wider path
• Sorting Blocks by color
• Supporting faster autonomous cycles

An optical sensor identified Block color before scoring so an unwanted Block could be rejected rather than scored.

The lower Center Goal funnel was built into the indexer. When the intake reversed for lower Goal scoring, the funnel narrowed the two-wide Block path into a single-file outlet before the Blocks reached the intake. This allowed the intake to eject them cleanly into the lower Center Goal.

The indexer, sorting path, and funnel were mechanically functional but needed more software integration and Block testing.

The outtake scored into the Long Goal and top Center Goal.

A simple pivot was selected instead of a four-bar because it:

• Reduced mechanical complexity
• Maintained more consistent Block compression
• Used fewer moving components
• Packaged more easily beneath the Long Goal
• Opened an additional scoring path when raised

The outtake used pneumatic flap states to control Block retention and scoring behavior.

Lower Center Goal scoring did not use the outtake. That action was completed by reversing the intake through the funnel built into the indexer.

The Gamma wing was included in the original robot architecture rather than added later.

It used separate pneumatic actions for lifting and actuating the mechanism, creating three primary states:

• Stowed
• Hover
• Descore

The stowed position remained below the Long Goal clearance limit. The hover state extended the wing over a scoring area, while the descore state repositioned it to remove Blocks.

Planning the wing from the beginning improved its mounting, packaging, deployment geometry, and structural support compared with adding the system after the robot was already complete.

The mechanism functioned, but its final durability, control, and strategic use still required more testing and driver practice.

The new Multi-Tool was one of the more successful Gamma mechanisms.

It used a polycarbonate plate that rotated freely on a shaft. A V-shaped cutout at the front guided the mechanism into the Match Loader.

The plate entered the Loader in a horizontal orientation. As it moved farther inside, contact with the Loader caused it to rotate into an angled position. This transformed the plate into a ramp that allowed the Match Loaded Blocks to fall directly toward the intake.

The passive rotation reduced the number of powered actions required and let the same piece adapt to the Match Loader geometry. Compared with the Luigi and Guido Multi-Tool, the new design interacted with the Loader more cleanly and worked well during testing.

The Gamma odometry pods moved vertically instead of rotating around a pivot.

Each pod:

• Slid linearly on rounded shafts
• Used bushings to guide the vertical movement
• Was spring-loaded downward against the field
• Used a custom magnetic encoder
• Provided tracking data to the goBILDA Pinpoint
• Packaged around the H-drive and low frame

The linear design reduced the swept volume required by a pivoting pod and maintained a direct vertical force into the field.

The pods were intended to be lifted pneumatically before parking, but that feature was not implemented. They remained spring-loaded downward in the completed robots.

The pneumatic layout controlled:

• Wing lift
• Wing actuation
• Indexer lower Goal funnel
• Outtake lift
• Outtake flap states
• Multi-Tool positioning

The design also reserved a function for lifting the odometry pods, but the pneumatic lifting mechanism was not implemented.

Sally and Cruz used the same core custom electronics stack developed for Luigi and Guido, except they did not use a Limelight.

The system included:

• goBILDA Pinpoint odometry computer
• Custom magnetic quadrature encoders
• Linear spring-loaded tracking-wheel pods
• Raspberry Pi communication coprocessor
• KUdos Raspberry Pi Expansion Hat
• Optical sensors for Block detection and color sorting
• Wireless hotspot for remote access
• Communication between custom devices and the V5 Brain

The Pinpoint calculated robot position and heading using its integrated IMU and the custom tracking-wheel encoders.

The Raspberry Pi acted as the communication bridge between the Pinpoint, other custom sensors, and the V5 Brain. No camera-based Limelight alignment or perception was used on Sally and Cruz.

Reliable localization was especially important for the H-drive because autonomous control needed to coordinate forward motion, lateral motion, and rotation.

I contributed system-level requirements, mechanical packaging, wiring, sensor mounting, and robot integration. I then used the completed localization and control foundation while programming the Skills and match autonomous routines.

I personally began programming Skills and match autonomous routines for Sally and Cruz.

The autonomous strategy included:

• Maximum-score Skills route planning
• Long Goal rush routines
• Center Goal rush routines
• Autonomous Win Point routines
• Color-sorted scoring
• Park Zone Block collection
• Double parking
• Wing hovering and descoring
• A combined top and bottom Center Goal descore concept
• Pinpoint-based forward, lateral, and rotational movement

The H-drive required a different motion approach than the tank-drive Luigi and Guido robots. The autonomous system needed to control three movement axes while maintaining reliable field position.

The physical robots were complete for only the final two weeks before Worlds. During that time, we still needed to:

• Finish the Skills route
• Complete the match autonomous routines
• Tune forward, lateral, and rotational movement
• Resolve drivetrain jitter
• Integrate every mechanism state
• Test color sorting
• Tune double parking
• Complete repeatability testing
• Practice recovery from failed actions

These were solvable development tasks, but the remaining schedule was not long enough to complete and validate them.

Sally and Cruz were physically completed two weeks before the 2026 VEX Robotics World Championship.

Their four-week development window had to include fabrication, assembly, wiring, plumbing, integration, programming, testing, tuning, and driver practice. Completing the builds halfway through that period left only two weeks for full-system development.

We brought Sally and Cruz to St. Louis alongside Luigi and Guido and continued working on them. Two days before Worlds, we determined that the robots had not received enough debugging, autonomous development, tuning, and repeatability testing.

Sally and Cruz had a substantially higher competitive ceiling than Luigi and Guido. Their architecture could support the movement and match strategies we believed were needed to contend for a World Championship.

Luigi and Guido were more limited, but they had already completed three competitions and were performing close to their ceiling. Their autonomous routines, controls, maintenance procedures, replacement parts, and failure modes were understood.

The decision was based on the remaining time and competition risk. With a longer development period, we could have continued working through the Gamma robots' issues and developing them toward their higher ceiling. For Worlds, we chose the proven robots that were ready.