Luigi & Guido · Push Back Competition Robots

The early Push Back season had very few nearby VEX-U competitions before the end of 2025. Our primary goal was to qualify for Worlds through the rule that invited the top five unqualified teams in the World Skills standings as of December 31.

This led us to design Luigi and Guido around the Skills Challenge instead of conventional early-season match play. The maximum score for either Driver or Autonomous Skills was 134 points, producing a theoretical combined maximum of 268 points.

Our highest scoring priorities were:

1. 1Park both robots
2. 2Clear both Match Loaders
3. 3Clear both Park Zones
4. 4Fill the Long Goal Control Zones
5. 5Fill both Center Goal Control Zones
6. 6Score the remaining Blocks

The robots later needed to transition from specialized Skills machines into competitive match-play robots without abandoning the architecture we had already designed and built.

Luigi and Guido were initially designed around Skills rather than conventional tournament play.

The robots used a high-capacity internal Block path so they could collect, carry, and score approximately 14 Blocks. A wide intake reduced alignment time, while the indexer and outtake controlled how Blocks moved through the robot.

The intended Skills workflow was:

1. 1Clear Blocks from the Park Zones
2. 2Collect loose Blocks from the field
3. 3Score into the Center Goals
4. 4Empty the Match Loaders
5. 5Score into the Long Goals
6. 6Fill the available Control Zones
7. 7Return both robots to the Park Zone

Placing the intake and outtake on opposite sides of each robot reduced cycle time. After collecting Blocks from a Match Loader, the robot could drive directly toward the Long Goal and score without turning around.

The architecture prioritized maximum scoring capacity and autonomous consistency. This created a strong Skills platform but introduced challenges when the robots later encountered defense, traffic, and direct robot interaction during tournament matches.

Alpha, Beta, and Beta+ were not separate robots. Luigi and Guido remained the same physical machines and were continuously upgraded based on testing, tournament feedback, and the team's changing strategy.

Alpha established the core architecture:

• Shared 24" and 15" mechanical design
• 8-motor motor-stack tank drivetrain
• Super-wide intake
• High-capacity indexer
• Pivoting outtake
• Long Goal aligner
• Multi-Tool for Loader and field interaction
• Last-minute deployable wings
• Odometry pods with custom magnetic encoders
• Raspberry Pi, Pinpoint, and Limelight integration
• Skills and match autonomous routines
• Provision for an indexer color-sorting chute that was not completed or used

The indexer was originally designed with a color-sorting chute, but the required sensor was never installed and the chute was not used in competition. Color sorting was later implemented at the outtake nozzle instead.

Alpha competed at Illini VURC Cornfield Clash 1. The robots ranked 3rd of 12 teams, earned the Design Award, finished as Tournament Finalists, placed 2nd in Robot Skills, and qualified for the VEX Robotics World Championship with a 9-4-0 overall record.

The event exposed several weaknesses:

• Indexer and outtake compression caused jamming
• Luigi's long aluminum intake shaft bent under load
• Several gold-filament parts were too brittle for repeated impacts
• The Raspberry Pi Python script was slow, laggy, and consumed too much memory
• Raspberry Pi crashes interrupted autonomous routines
• Interrupted routines reduced scoring, created inconsistencies, and could cause rule violations
• V5 Distance Sensors provided useful wall-distance measurements but unreliable heading estimates
• The Match Loader alignment process needed greater consistency
• The wings were added one day before the event and were poorly integrated
• The wing structure and mounting could not reliably resist impacts acting through the long lever arm

Beta retained the core Alpha architecture while improving reliability, autonomous consistency, and match-play capability.

Major improvements included:

• Completely redesigned the deployable wings
• Reworked the wing geometry, mounting, deployment, and hard stops
• Used sandwiched PETG and polycarbonate wing structures
• Replaced the bent aluminum intake shaft with a steel shaft
• Replaced broken gold-filament components with stronger white-filament parts
• Continued using some gold parts while preparing white replacement parts in case they failed
• Reinforced the outtake backing and drivetrain components
• Refined the intake and indexer transitions to reduce jamming
• Moved color sorting from the unused indexer chute to the outtake nozzle
• Switched the Raspberry Pi backend from Python to C++
• Reduced memory usage, lag, and autonomous interruptions
• Used the Pinpoint IMU for heading
• Continued using V5 Distance Sensors to measure distance from field walls
• Used wall-distance measurements to reset positional estimates during autonomous routines
• Added reliable Limelight Match Loader alignment
• Improved Raspberry Pi mounting, power reliability, and troubleshooting access
• Refined driver-control macros and autonomous routines
• Continued improving the Multi-Tool for Loader clearing and field interaction

The change from Python to C++ was especially important because the purpose of the custom electronics stack was to provide precise and repeatable autonomous performance. Crashes and lag directly undermined that goal by interrupting routines, reducing scores, creating field-position errors, and occasionally causing violations.

Beta competed at Gear Slingers Open 2026 and California Baptist University VEXU Open 2026. The robots ranked 3rd at both events, earned two Excellence Awards, finished as Tournament Semi-Finalists and Tournament Finalists, and placed 2nd in Robot Skills at both competitions.

Beta+ was the final Worlds configuration of Luigi and Guido.

After California Baptist University VEXU Open 2026, the team's primary development effort shifted to Sally and Cruz. Luigi and Guido remained fully assembled and ready as a backup, but we did not expect to use them at Worlds.

Sally and Cruz were built, and autonomous programming had begun. However, we did not have enough time to complete the Skills and match routines, resolve the remaining mechanical and software bugs, or properly tune the new robots. We brought all four robots to St. Louis and continued working before deciding two days before Worlds to compete with Luigi and Guido.

Because the decision was made so late, the most significant Beta+ mechanical change was completed at the event. On the morning of Worlds Day 1, I designed, tested, manufactured, and installed a new wing aligner.

The Beta+ wing system included:

• A redesigned four-bar wing aligner that deployed with the wing
• Rollers at the end of the aligner that glided against the Long Goal
• Geometry that guided the robot into a repeatable scoring position
• A structure designed to withstand robot impacts acting across a long lever arm
• PAHT-CF wing mounts and aligner components
• Sandwiched PETG and polycarbonate wing structures
• Stronger mounting than the PETG and polycarbonate hybrid versions that repeatedly broke during Beta

There was no wing aligner on Beta. The complete four-bar aligner was added for Beta+ after the decision to use Luigi and Guido at Worlds.

Additional Beta+ work included:

• Retuned Skills autonomous routines
• Retuned match autonomous routines
• Inspected and repaired the existing Beta mechanisms
• Prepared replacement parts and maintenance procedures
• Adjusted match strategy around the proven capabilities of Luigi and Guido

Beta+ competed at the 2026 VEX Robotics World Championship, where the robots ranked 7th of 41 teams in the Opportunity Division with a 9-1-0 qualification record, finished with a 9-2-0 overall record, placed 19th in Robot Skills, and earned the Innovate Award.

Luigi and Guido shared the same core subsystem architecture across their 24" and 15" configurations. Each subsystem was packaged around a continuous Block path and continuously revised through Alpha, Beta, and Beta+.

• Used a tank drivetrain for predictable Skills pathing and pushing ability
• Powered each robot with 8 drivetrain motors
• Ran the drivetrain at approximately 512 rpm
• Used a motor-stack arrangement to package the motors compactly
• Integrated front and side ramps to help enter the Park Zone
• Used odometry pods for parallel and perpendicular position tracking
• Added custom motor gears with external shaft collars for easier inspection and service
• Designed removable motor caps so damaged or overheated motors could be replaced
• Used the drivetrain as a stable base for the intake, indexer, outtake, Multi-Tool, electronics, and pneumatics

The drivetrain worked well during Skills but had limited acceleration and maneuverability during aggressive match play. Defensive contact could stall or pin the robots, and their height prevented them from driving beneath the Long Goal.

• Spanned most of the robot's width to reduce alignment time
• Collected Blocks from the floor and directed them into the indexer
• Used funnels to center Blocks before they entered the internal scoring path
• Allowed the robot to collect scattered Blocks while following Skills routes
• Used replaceable printed components and rollers
• Received stronger shafts and revised plates after Alpha testing
• Replaced Luigi's long aluminum intake shaft with steel after it bent
• Integrated with the Multi-Tool and starting configuration

The intake was positioned opposite the outtake, allowing the robot to collect Blocks from a Match Loader and drive directly to the Long Goal without turning around. This reduced the time between loading and scoring.

• Carried approximately 14 Blocks through a continuous internal path
• Used multiple powered rollers to move Blocks through the robot
• Used gates and software states to control Block movement
• Coordinated with the intake and outtake through the robot's state-machine system
• Used replaceable backings, idlers, spacers, and transition parts
• Included an early color-sorting chute concept that was not completed or used
• Later sorted unwanted Blocks at the outtake nozzle instead

The original architecture considered storing at least 7 Blocks of each color and independently controlling both groups. That concept was not implemented. Although the indexer included a mechanism for a color-sorting chute, the sensor was never installed and the chute was not used.

Later in the season, color sorting was moved to the outtake nozzle. This provided a more direct way to reject an unwanted Block before it was scored without requiring separate internal storage paths.

Compression between the rollers and backings caused recurring jams. Multiple backing, spacer, roller, and transition revisions were developed throughout Alpha and Beta to improve Block flow.

• Scored Blocks into the Long Goals
• Changed position to reach the top Center Goal
• Used a pivoting architecture to remain compact
• Used a movable flap to control the scoring angle
• Used the flap to prevent Blocks from falling out while the robot was intaking
• Used goal aligners to position the robot against the Long Goal
• Used flexible tubing and replaceable backing components
• Coordinated with the indexer through software states
• Later incorporated color sorting at the outtake nozzle

Moving approximately 14 Blocks through the robot generated significant force against the outtake. Without the movable flap, Blocks could be pushed out of the robot while the intake and indexer were running. Closing the flap retained the Blocks during collection, while repositioning it allowed the robot to score into different Goals.

The simple pivoting architecture provided multiple scoring positions without requiring a more complex linkage. It later influenced parts of the Sally and Cruz architecture.

The Multi-Tool was a deployable subsystem used for several game actions:

• Cleared Blocks from the Match Loaders
• Scraped Blocks from the Park Zones
• Assisted with scoring into the lower Center Goal
• Helped position and control Blocks around the robot
• Acted as part of the intake's starting restraint
• Used ramps shaped like an F1 drag-reduction-system spoiler as part of the Cars theme

The mechanism was continuously revised through new ramps, side plates, spacers, pivots, hard stops, and zeroing methods.

The Alpha wings were added one day before Illini VURC Cornfield Clash 1. They proved that the robots could benefit from wings during match play, but the rushed design was poorly integrated and lacked the strength and consistency needed for competition.

Beta introduced a complete wing redesign rather than a minor revision. The new wings improved deployment geometry, hard stops, field coverage, and mounting. They used sandwiched PETG and polycarbonate structures, but repeated impacts still damaged the wings and their mounts.

The wings were used to:

• Move groups of Blocks
• Descore Blocks from Long Goal areas
• Defend scoring positions
• Control space around the Goals
• Assist with match-specific strategies

Both the wings and their mounts experienced high loads because impacts occurred far from the pivot, creating a large lever arm. This made strength at the pivot, mounts, and deployment structure critical.

For Beta+, the wing mounting components were rebuilt using PAHT-CF, while the wing structure retained the sandwiched PETG and polycarbonate construction. The new PAHT-CF four-bar wing aligner deployed with the wing, used rollers to glide against the Long Goal, and was designed to resist robot impacts.

Luigi and Guido were designed around a custom sensor and localization stack that could support VEX-U autonomous routines and future VEX AI development.

The system included:

• goBILDA Pinpoint odometry computer
• Custom magnetic quadrature encoders
• Spring-loaded odometry tracking wheels
• Limelight 3A smart vision camera
• Raspberry Pi communication coprocessor
• V5 Distance Sensors for wall-distance measurements
• Optical sensors for Block detection and outtake color sorting
• Wireless hotspot for troubleshooting and updates
• Communication between custom hardware and the V5 Brain
• Expansion capacity for additional cameras and sensors

The Pinpoint independently calculated the robot's position using its integrated IMU and the custom tracking-wheel encoders. The Pinpoint IMU became the primary source for robot heading because it was more consistent than estimating orientation from the V5 Distance Sensors.

The Distance Sensors were still useful for measuring positional distance from known field walls. Autonomous routines could use these readings to reset or correct the robot's estimated field position.

The Limelight ran vision processing on its own hardware. A neural network detected the yellow cap on the Match Loader, allowing the robots to automatically align despite field-position and lighting variation.

The Raspberry Pi served as the communication bridge between the custom sensors and the V5 Brain. Alpha initially used a Python script, but it consumed too much memory and produced slow, laggy performance. Crashes interrupted autonomous routines, reduced scoring, created inconsistent movement, and could cause violations.

For Beta, the Raspberry Pi backend was rewritten in C++. This improved memory use, speed, and stability, allowing the electronics stack to better fulfill its purpose of providing precise and consistent autonomous performance.

I contributed concepts and system requirements for the electronics stack, designed its mechanical packaging, soldered and assembled the encoder hardware, installed the sensors, and integrated the hardware into the robots. I then used the completed localization and control backend to develop and tune the autonomous routines.

Serving as Drive Coach also influenced autonomous development. I planned Skills routes, selected match autonomous routines, coordinated starting positions with alliance partners, and used competition results to identify paths and actions that needed additional tuning.

The autonomous routines used Pinpoint odometry, Limelight alignment, Distance Sensor wall resets, optical sensors, and the robots' subsystem state machines.

I worked on:

• Skills route planning
• Match autonomous pathing
• Autonomous scoring order
• Pinpoint-based field paths
• Pinpoint IMU heading control
• Distance Sensor field-wall resets
• Match Loader alignment
• Long Goal alignment
• Parking routines
• Driver-assist macros
• Event-specific counter-autonomous routines
• On-field autonomous tuning

The software backend handled localization, communication, and subsystem states. I used that foundation to create and tune the actual routes and scoring sequences needed for competition.

At Gear Slingers Open 2026, the Pinpoint and Limelight systems became extremely consistent. The robots reached a 105-point Driver Skills score and performed reliable Match Loader alignment.

At California Baptist University VEXU Open 2026, the combined Skills score reached 162 points with 112 Driver Skills points and 50 Autonomous Skills points.

Sally and Cruz were built and had begun autonomous development before Worlds, but there was not enough time to complete and tune their Skills and match routines. Returning to Luigi and Guido allowed us to use established autonomous routines that only needed to be adjusted for the Beta+ configuration and Worlds fields.

Sarge and Fillmore were two simplified Autonomous Skills robots built a few days before California Baptist University VEXU Open 2026. Their names continued the season's Cars theme.

We developed them after missing the Skills Champion award by 2 points at each of our first two tournaments. Our autonomous score at both events came from double parking for 30 points. Adding a cleared Park Zone would have increased that routine to 35 points, which would have been enough to win Skills Champion at both events by 3 points.

Sarge and Fillmore did not have drivetrains. They were designed only to complete the minimum actions required for a reliable 35-point Autonomous Skills score:

• Sarge used a pneumatic mechanism to push the Blocks out of the starting Park Zone for 5 points
• Fillmore spun a motor so it registered as a moving robot
• Sarge remained parked for 15 points
• Fillmore remained parked for 15 points
• Total score: 35 points

The concept intentionally avoided autonomous pathing, localization, and complex scoring mechanisms. Their purpose was to establish a dependable 35-point Skills baseline in case Luigi and Guido's longer autonomous routine experienced mechanical, electronics, or localization problems.

We used Sarge and Fillmore for our first Autonomous Skills run at California Baptist University VEXU Open 2026, where they successfully established the 35-point baseline. Luigi and Guido then exceeded that score with 42-point and 50-point Autonomous Skills runs, so Sarge and Fillmore were not used for our final combined score.

Although the backup robots were ultimately unnecessary, they gave the team a low-risk contingency and ensured that an autonomous failure from Luigi and Guido would not leave us below the score that had cost us Skills Champion at the previous two events.

The robots were mechanically and electronically designed with VEX AI in mind from the beginning of the season.

This influenced the decision to include:

• An external localization computer
• Camera-based field alignment
• A Raspberry Pi coprocessor
• Custom encoders and sensor interfaces
• A modular communication architecture
• Wireless troubleshooting and configuration
• Expandability for additional vision cameras and sensors

Luigi and Guido did not compete in VEX AI. After completing the VEX-U World Championship, the team chose to take a break instead of preparing for another major event. No VEX AI-specific mechanical, electrical, or software testing was completed.

After California Baptist University VEXU Open 2026, I led the development of Sally and Cruz, a replacement robot pair intended to address the limitations discovered through Luigi and Guido.

The new architecture focused on:

• Driving beneath the Long Goal
• Holonomic movement
• Faster acceleration
• More reliable double parking
• Lower Block compression
• Stronger and pre-planned wings
• Better defensive maneuverability
• Improved match play
• Continued use of the custom electronics stack

Sally and Cruz were fully built, and the team had begun programming their autonomous routines. However, the project fell behind during integration and testing. We did not have enough time to resolve the remaining bugs, finish the Skills and match autonomous routines, and properly tune both robots.

We kept Luigi and Guido built, functional, and competition-ready while developing Sally and Cruz. All four robots were brought to St. Louis so we could continue working while preserving a reliable backup option.

Two days before Worlds, while already in St. Louis, we decided that Sally and Cruz were not ready for competition. We returned to Luigi and Guido because their mechanisms, autonomous routines, replacement parts, and maintenance requirements were already understood.

Sally and Cruz are covered in greater detail as a separate portfolio project.