Rafiki & Simba · High Stakes Competition Robots

High Stakes rewarded Ring scoring, Mobile Goal control, and climbing. The highest-value repeatable objective available to a VEX-U team was a Tier 3 Buddy Climb, which doubled Simba's 12-point Tier 3 Climb to 24 points.

Following the team's principle that strategy dictates design, our goal was to create a system that could establish a high score before opponents had a chance to interfere. Rafiki and Simba needed to dock, use opponent Rings to complete Rafiki's Tier 1 Climb, lift Simba to Tier 3, and score on the High Stake.

Together, these actions could score 39 points using a single scored Ring:

The system also needed to declimb, control Mobile Goals and Positive Corners during driver control, deny opponent Rings, and repeat the climb at the end of the match.

The original concept, nicknamed the "Wall-E Backpack," used an extremely lightweight mini-bot that could fit within the space occupied by two Rings. Rafiki would lift the mini-bot while using the same manipulator for Ring scoring.

We later determined that the robots needed to remain connected throughout the match so opponents could not prevent Rafiki from retrieving the smaller robot before climbing. Once the two-Ring packaging constraint was removed, the smaller robot could support its own arm and claw. This became Simba.

The final strategy used both robots as one combined system:

1. 01Dock Simba to Rafiki
2. 02Intake two opponent Rings beneath Rafiki
3. 03Contact the Ladder
4. 04Extend Rafiki's platform lift and sit on the two Rings for a Tier 1 Climb
5. 05Raise Simba with the double reverse 4-bar for a Tier 3 Buddy Climb
6. 06Use Simba's claw to score a Ring on the High Stake
7. 07Declimb during driver control when additional field interaction was required
8. 08Protect the Positive Corner, control Mobile Goals, and deny opponent Rings
9. 09Return to the Ladder and climb again before the match ended

As drive coach, one recurring in-match tactic was to delay leaving the Positive Corner until roughly 15 to 25 seconds remained, baiting opponents into committing to a tempting open Goal before we climbed, trading a little Goal value for a safer, less contested climb.

High Stakes defined four horizontal planes above the field: the ground, and Ladder Tiers 1, 2, and 3. A Robot could only touch or cross two of these planes at once; breaking a third was illegal. The game manual itself calls the resulting numbers "effective height limits," not hard limits: they aren't heights written directly into the rules, but the tallest a Robot's actual geometry can reach without touching or crossing that third plane.

Grounded, Rafiki was already touching the ground plane, which used up one of its two planes and left it only Ladder Tier 1 to cross. Ladder Tier 2 itself sits at 32", so with the ground and Tier 1 already accounting for both allowed planes, Rafiki could not touch or cross Tier 2, giving it an effective height limit of 32". I designed the lift with a software-controlled floor mode that capped its height at 27-7/8", leaving 4-1/8" of margin beneath that limit.

To free up a plane, Rafiki had to stop touching the ground entirely rather than just get taller. Using its Ring Intake, it pulled two Rings beneath its chassis and contacted the Ladder, then four pneumatic cylinders extended the Platform Lift, raising the robot fully onto the Rings. V5RC Q&A 2093 confirmed that resting on Rings, rather than the floor, did not prevent a Robot from being considered Climbed, so once Rafiki was supported by the Rings, off the ground, and touching the Ladder, it qualified as a Tier 1 Climb.

Climbed, Rafiki was no longer touching the ground plane, so its two-plane budget shifted upward to Ladder Tiers 1 and 2, raising its effective height limit from 32" to 46". In climbed mode, the lift was software-limited to a maximum height of 42-3/16", leaving 3-13/16" of margin beneath that limit. The software used separate limits for the floor and climbed states so the same mechanical lift could operate legally throughout the sequence.

Reaching Tier 3, where the High Stake sits, needed a third plane broken, which Rafiki's own two-plane budget could not legally do. This is where docking two separate Robots mattered: Simba is legally a distinct Robot from Rafiki with its own independent two-plane budget, so while Rafiki's structure stayed within Ladder Tiers 1 and 2, Simba used its own separate allowance to cross into Tier 3 and score the High Stake, completing the Tier 3 Buddy Climb.

Ring possession worked the same way. Simba could legally hold up to two Rings of its own, and separately, Rafiki possessed the two Rings it stood on for the Platform Lift the entire time, even while lifting Simba: V5RC Q&A 2026 confirmed that carrying Simba did not make Rafiki possess the Rings Simba held, since Ring possession is not transitive between docked Robots.

Repeated testing of both the floor and climbed limits, in manual and climb-macro control, confirmed the robot held roughly 4" of margin beneath each plane while Simba still reached solidly into Tier 3 for the full 24-point Buddy Climb.

The climb sequence was:

1. 01Dock Simba to Rafiki
2. 02Intake two Rings beneath Rafiki
3. 03Contact the Ladder using the Ladder arm or flexible contact brushes
4. 04Extend the pneumatic Platform Lift to get off the ground and onto the Rings for a Tier 1 Climb
5. 05Switch the double reverse 4-bar to its climbed software limit
6. 06Raise Simba to a Tier 3 Buddy Climb
7. 07Have Simba cross the top most plane and score on the High Stake using Simba's claw

This approach did not bypass the expansion rule. It used two legally separate Robots, each staying within its own two-plane budget, to reach a combined height that neither robot could have legally reached alone.

Alpha, Beta, and Gamma were internal revision names for the season-long development of Rafiki and Simba rather than separate project entries.

Rafiki retained the same core robot structure and was continuously modified throughout the season, moving through Alpha, Beta, and Gamma. Simba had no separate Beta iteration: it kept its Alpha design unchanged through the Beta era, docked to the upgraded Rafiki, before being completely redesigned for Gamma once Alpha and Beta had proven out the geometry, subsystem interfaces, and strategy requirements.

Alpha established the core architecture used throughout the season:

• 6-motor double reverse 4-bar
• 8-motor Rafiki drivetrain
• Pneumatic docking between Rafiki and Simba
• Pneumatic platform lift
• Simba virtual belted 4-bar
• Two-Ring claw
• Mobile Goal clamp
• Autonomous climb and High Stake sequence

Building Alpha also surfaced a major drivetrain friction problem, traced to the wheel hubs and axle shafts (see Drivetrain below).

Alpha competed at the West Michigan Holiday Tournament and RiverBots III. These events exposed weaknesses in Simba's printed base and arm pivot, the docking mechanism, Rafiki's lift rigidity, Ladder contact, and vertical-expansion stability. At West Michigan specifically, Simba's arm died mid-event and left the pair stuck climbed for half of a driver-control period, a concrete reliability failure that fed directly into the Beta priorities.

Beta was a Rafiki-only revision; Simba was not redesigned and continued docking to Rafiki in its original Alpha form. Beta retained the same overall architecture and continued using a 6-motor double reverse 4-bar with an 8-motor drivetrain.

Major improvements included:

• Ratchet system to hold the double reverse 4-bar in position
• Reinforced lift plates, motor caps, and gear spacing
• Improved hot-swap access for lift motors and gears
• More reliable Ladder contact mechanism, replacing the pivoting arm with a rack-and-pinion linear slide
• 0.5" stroke to 1" stroke pneumatic cylinders on both the platform lift and Mobile Goal clamp
• Side and vertical skirts for more climb alignment opportunities
• Improved docking durability and alignment
• Faster declimb sequence
• Greater serviceability between matches

Beta competed at the Purdue Slam and Jam qualifier, where a detailed six-match post-mortem shaped the Gamma priorities. Recurring issues included the double reverse 4-bar's gears chipping while declimbing after autonomous (dropping a Tier 3 to a Tier 2) and the 15" claw skipping during a High Stake attempt. The team still won several matches, including one on an opponent rules disqualification, but the pattern of mechanical faults made clear that Gamma had to prioritize durability.

Gamma was the final Worlds configuration. Rafiki received another major revision, while Simba, still running its original Alpha design, was completely redesigned around the architecture that Alpha and Beta had proven out.

The redesign was easier to complete because Simba's core requirements were already understood. It still needed to dock with Rafiki, use a virtual belted four-bar, manipulate Rings, score on the High Stake, and function as the lifted robot during the Tier 3 Buddy Climb. Instead of developing a new concept, I could focus the redesign on structure, packaging, reliability, weight, and serviceability.

Major Rafiki improvements included:

• Reduced the double reverse 4-bar from 6 motors to 4 motors
• Added 2 x 10 lbf gas struts to assist the double reverse 4-bar
• Reallocated the removed lift motors into the drivetrain, settling on a 10-motor layout after a 12-motor version could not clear the double reverse 4-bar without giving up motor hot-swap
• Made the double reverse 4-bar motors hot-swappable within one to two minutes between matches
• Reinforced the lift with polycarbonate plates and improved gear support
• Improved the docking geometry and mounting interfaces
• Added more robust Ladder contact and climb-alignment features
• Tripled pneumatic storage capacity
• Refined pneumatic, electrical, and mechanical packaging
• Improved replacement-part access and between-match maintenance

Major Simba redesign changes included:

• Completely rebuilt the robot while retaining its established functional architecture
• Replaced the reinforced printed base with a pocketed 1/4" aluminum baseplate
• Reduced the baseplate weight from approximately 1.225 lbs to 0.597 lbs, a 49% reduction
• Repackaged the drivetrain, Robot Brain, Battery, Radios, docking hardware, and arm structure
• Redesigned the arm pivot and mounting structure
• Revised the virtual belted four-bar and claw power transmission
• Improved the docking interface using geometry developed through the Alpha and Beta revisions
• Increased structural rigidity around the arm, drivetrain, and docking loads
• Improved access to motors, electronics, fasteners, and replaceable components
• Created cleaner and more repeatable mounting interfaces between subsystems
• Created a modular mounting system for the claw to be able to swap out different claw types (roller claw and simple claw)

Gamma competed at VEX-U Worlds and VEX AI Worlds.

Rafiki served as the primary drive platform and supported the combined climb, docking, and Mobile Goal control systems.

• 8 motors through Alpha and Beta, reallocated to 10 motors for Gamma
• Geared 36T to 48T, from 600 rpm cartridges down to about 450 rpm
• Ran on 4 omni and 4 traction wheels, 3.25" diameter
• Carried custom tracking-wheel odometry pods for the entire season, feeding wheel-based position data to the autonomous routines
• Stacked the drive motors vertically at the back of the robot instead of the usual layout, doing double duty as a rear counterweight against Simba's weight when lifted and freeing up the bottom of the chassis for the Ring Fence, Ring Intake, and Platform Lift

Building Alpha surfaced a major drivetrain friction problem, with the motors each drawing around 8 watts. I traced it to support material trapped inside the printed wheel hubs, which misaligned the bearings, and to the 1/4" round axle shafts digging into the printed drivetrain rails. Reprinting the hubs with a sacrificial bridging layer and adding counterbored metal washers so the shafts could not dig in dropped the draw to roughly 0.1 to 0.5 watts per motor.

• Lifted Simba to a Tier 3 Buddy Climb
• Used a 12:84 gear reduction, from 100 rpm motor cartridges down to about 14 rpm
• Used 6 motors during Alpha and Beta
• Used 4 motors and 2 x 10 lbf gas struts during Gamma
• Used 1/2" x 1/2" x 1/16" wall box-tube arms for rigidity and compact packaging
• Ran the linkage joints on 1/4" aluminum standoff dead axles
• Used polycarbonate outer plates, 3D printed motor cap, inner polycarbonate reinforcement plates, and a polycarbonate gear-tensioning plate
• Allowed the lift motors and gears to be serviced between matches

Under Simba's full weight, the large 84-tooth lift gears skipped teeth because their 1/4" steel shafts bowed apart. I resolved this with four polycarbonate gear-tensioner plates that held the gears at a fixed center-to-center distance, keeping the shafts from bending apart under load, combined with software slew that ramped the lift's acceleration and deceleration, which together eliminated almost all of the skipping.

Holding the lift also went through a full redesign. Alpha used a sliding pneumatic lock that drove 3D printed teeth into the 84-tooth gears, adapted from KUdos's previous Over Under climber PTO, but it jammed during matches, which risked leaving the lift stuck in a raised position for an entire match. For Beta I replaced it with a true one-way ratchet, a washer that catches the gear, mounted on each side of the lift and released by a pneumatic cylinder, so the motors could coast or brake instead of holding and hold Tier 3 at the end of the match. Releasing it had its own quirk: even after the cylinder retracted, the ratchet stayed bound under the lift's load, so declimbing early on took about 15 seconds until the release logic was changed to nudge the lift rather than just drop the pneumatics. In software, the lift's hold strategy evolved in parallel, from a simple hold, to slew-limited motion, to a reduced-current hold, and finally the pneumatic brake.

The ratchet washer itself went through a material study. Beta used steel washers, which held up but could damage the DR4B's gear teeth on impact. Gamma switched to 3D printed washers, which stopped the gear damage but would occasionally break themselves, so we replaced them after every match as a precaution. By VEX AI Worlds we had moved to polycarbonate washers, which held up without breaking and without damaging the gear teeth.

For Gamma, I sized the assist gas struts with a static free-body-diagram analysis of the roughly 9 lb moving lift structure, which called for about 41 lbf in the extended position and 28 lbf retracted. I deliberately selected weaker approximately 10 lbf struts so they assisted the motors without forcing the lift upward when the robot was disabled.

• Collected Rings from the floor and fed them beneath the chassis for the platform lift
• Used 2" flex wheels driven at 200 rpm
• Tested at 100 rpm and 200 rpm, and settled on 200 rpm for enough intaking torque

• Established Ladder contact so Rafiki could legally enter a climbed state
• Used exclusively for the autonomous climb, where a precise, repeatable approach could be programmed without opponents interfering
• Driven by a single 100 rpm motor
• Used a pivoting arm during Alpha
• Replaced the pivoting arm with a rack-and-pinion linear slide for Beta for more consistent Ladder contact

• Flexible pneumatic-tubing contact surfaces added along the sides of the drivetrain and the sides of the top of the DR4B base pieces
• Drivetrain-side skirts contacted the Ladder base; DR4B-side skirts contacted the Tier 1 horizontal bar
• Used exclusively during driver control, where lining up the precise Ladder Arm was much harder under active defense
• Gave Rafiki a wide, forgiving range of positions from which it could still touch the Ladder for a second climb late in the match
• Once climbed, Rafiki could no longer be legally defended against, so a faster, more forgiving second climb meaningfully reduced risk late in a match

• Used four 0.75" bore pneumatic pancake cylinders to raise Rafiki off the floor, 0.5" stroke during Alpha and 1" stroke from Beta on
• Supported the combined weight of Rafiki and Simba (39 lbs)
• Trapped two Rings beneath the chassis using a Ring fence
• Used a concave platform shape to improve Ring alignment
• Completed a Tier 1 Climb while Rafiki contacted the Ladder
• Reduced the climb's dependence on precise Ladder alignment

The Alpha 0.5" stroke was not enough lift on uneven field tiles: even fully extended, Rafiki sometimes still brushed the ground. The 1" stroke used from Beta on gave enough extra travel to reliably clear the ground regardless of tile unevenness.

• Connected Simba to Rafiki at the beginning of the match
• Used two 0.5" stroke, 0.75" bore pneumatic pancake cylinders producing approximately 88 lbf at 100 PSI
• Used matching positive and negative alignment geometry between the robots
• Guided the docking hardware into chamfered locating features
• Kept both robots connected during scoring, driving, climbing, and declimbing
• Developed through multiple revisions to improve alignment and prevent printed-part damage

• Used two 0.75" bore pneumatic pancake cylinders, 0.5" stroke during Alpha and 1" stroke from Beta on
• Produced approximately 88 lbf of clamping force at 100 PSI
• Captured the lower edge of the Mobile Goal against fixed standoffs
• Allowed Rafiki to control Mobile Goals for scoring and point denial
• Used a wide engagement area to reduce alignment sensitivity

The longer 1" stroke from Beta on also widened the clamp's grab range, making it easier to capture a Mobile Goal that wasn't perfectly aligned. Because Rafiki wasn't scoring Rings directly onto the Mobile Goal it held, unlike many other teams' designs, the clamp had no reason to grip tightly or precisely; it could be built as wide and forgiving as possible purely to grab a Mobile Goal easily and hold it for control and point denial.

Simba served as the Ring manipulator and the elevated portion of the Tier 3 Buddy Climb. It had no separate Beta version, competing through the Beta era in its original Alpha form; the Gamma version was then a complete mechanical redesign that preserved the role, docking relationship, virtual four-bar architecture, and scoring capabilities that design had already proven.

• Used a printed and steel bar reinforced structural base throughout Alpha, unchanged into the Beta era
• Repeated cracking in the printed base helped define the Gamma redesign requirements
• Replaced the previous base with a cnc'ed billet 1/4" aluminum baseplate for Gamma
• Reduced the baseplate weight from approximately 1.225 lbs to 0.6 lbs, though the redesign briefly grew to about 9 lbs mid-development before targeted trims brought it back down
• Repackaged the Robot Brain, Battery, Radios, drivetrain, docking interface, and arm pivot
• Used tapped and clearance holes for repeatable subsystem mounting
• Integrated the docking interface directly into the redesigned structure
• Improved rigidity, service access, and replacement-part consistency

• Pivoted on 2 x 11W motors with 100 rpm cartridges
• Used two 105T HTD 9 mm timing belts
• Maintained the claw's orientation while the arm moved
• Originally used a plain 3D printed bore at the arm pivot, which broke at West Michigan
• Replaced with a printed bore with a gear cutout to press-fit a 12T steel gear insert, so the bore itself no longer took the load and wore out; this held up through RiverBots, Worlds, and the rest of the season
• Ran the claw pulleys on 3/8" aluminum standoff dead axles
• Allowed Simba to manipulate Rings while docked to Rafiki

• Held up to two Rings
• Scored on Mobile Goals, Wall Stakes, the Alliance Stake, and the High Stake
• Used a single motor, 5.5W at 200 rpm during Alpha and upgraded to 11W at 100 rpm for Gamma
• Used herringbone gears for Gamma to reduce backlash
• Used replaceable 1/8" polycarbonate claw plates to play with claw geometry
• Allowed claw geometries to be changed by removing six screws

The climb relied on a detailed interpretation of High Stakes scoring and expansion rules.

• V5RC Q&A 2093 confirmed that a robot could achieve a Tier 1 Climb while sitting on Rings, provided it no longer contacted the floor and met the other climb requirements
• V5RC Q&A 2026 confirmed that Rafiki carrying Simba did not make Rafiki possess the Rings held by Simba and that possession was not transitive
• Rafiki used Ladder contact mechanisms and flexible skirts to establish climb contact
• The lift used software limits to prevent the robots from breaking more vertical-expansion planes than allowed
• Rafiki's lift height was limited differently while on the floor and while supported by the platform lift
• The ratchet held the double reverse 4-bar extended without relying only on motor torque
• The buddy climb also worked in reverse: during a West Michigan match, the 15" ended up contacting the Ladder for its own regular climb points while the 24" scored the Buddy Climb points, the opposite of the robots' usual roles, and both still scored legally

For Gamma, I developed a roller claw intended to collect one Ring, two stacked Rings, or a second Ring after the first had already been captured.

The design used 1" VexPro grip tube glued to 3D printed hubs with integrated sprockets, running on R188ZZ bearings at roughly 342 rpm, with the claw held open at the start of the match and deployed by spinning a roller to release a string. The hardest problem was ring stacking: a second intaked Ring would hit the first Ring before reaching the rollers, pushing it away instead of stacking it, which I worked to solve by extending the rollers outward and adding a funnel plate that lifted the first Ring higher. That reshaping moved the roller center-to-center from 3.869" to 4.710" and the roller minimum distance from 2.775" to 3.337", and changed the stowed Ring from a 0.125" compression to a 0.125" gap so the first Ring could move freely. Although the claw could collect Rings during testing, its deployment was not reliable enough before VEX Worlds.

With approximately one week remaining before the event, the roller claw was abandoned and Simba returned to the proven claw. This protected the reliability of the complete robot system instead of introducing an unfinished mechanism at Worlds.

I did not write the general subsystem or drivetrain code, which the team's programmers handled, but I worked directly on the autonomous paths and routes themselves, most importantly for VEX AI Worlds, where both robots had to operate the entire match with no human drivers. The routines ran on custom drivetrain odometry pods that Rafiki carried the entire season, giving the code accurate wheel-based position feedback. I also proposed several software concepts that teammates then implemented:

• Current-based homing for the double reverse 4-bar: press the arm down until motor current spikes to confirm it has bottomed out, so the lift always starts from a known, consistent position
• A points-based Skills autonomous progression, built up match by match from a 30-point baseline toward 65 points as Mobile Goals, Rings, and the High Stake were added into the route
• A shared timer to synchronize the 24" and 15" autonomous routines, after relying on Rafiki's own inconsistent timing made Simba's actions unreliable

By the final week before VEX AI Worlds, the fully autonomous climb-and-score routines were timing between 6.48 and 8.11 seconds per run in testing.

I contributed to the documentation, graphics, and media used to communicate the Rafiki and Simba project throughout the season. To capture build entries more reliably, the team also built a custom Discord submission bot that logged notebook entries directly from our meeting threads.

This work included:

• Engineering notebook entries
• Robot reveal videos
• Team and robot brochures
• Recruitment flyers
• Sponsor packets
• Team shirts
• Competition graphics
• Technical diagrams and CAD renders
• Photos and media used for judging, sponsors, outreach, and recruitment

The robot media received more than 45,000 YouTube views. Rafiki and Simba were also presented on the VEX Worlds Finals stage in front of a live audience of more than 8,000 people.