How your robot picks up and scores Override's pins and cups under the 1-pin-and-1-cup possession limit. Read Lift Mechanisms for the lift this manipulator attaches to, and Mechanism Concept Sprint if you have not yet brainstormed alternatives.
🏁 Override 2026-27 specific🧰 Most consequential mechanism choice this season⚙ Build decision — commit before kickoff
This page was rewritten on April 26, 2026 after the official Override game reveal video was released. The game has two distinct manipulables (pins and cups) and a strict possession limit (1 of each at a time). The earlier version of this page assumed a different element design and was wrong — the rewrite reflects the actual confirmed game.
Status of this guide: the reveal video confirmed the game elements, possession limit, and scoring rules used here. The official manual drops Monday April 27, 2026 — specific dimensions, masses, and PTO rules may refine some recommendations. Re-read this page after the manual is published.
The Manipulation Problem in One Paragraph
Override gives you 119 scoring objects across two types: pins (small, 4 color combinations, 63 total) and cups (hollow hourglass shape, one opaque side and one transparent side, 56 total). To score, you place a pin on a goal, then stack a cup on top of the pin. The cup's orientation matters — if a pin sits in the opaque side of a cup, that color doesn't count. Your robot can hold 1 cup AND 1 pin at any given time, no more. So your manipulator has to handle two different objects with different geometries, in a specific stacking order, with an orientation requirement on the cup.
This is not a simple gripper problem. The 1-of-each possession limit means you can't build an accumulator that holds multiple pins. The two-object handoff means a single all-purpose grip might be slower than two specialized grips. The cup orientation requirement means you may need a sensor and a rotation mechanism to make sure the cup is the right way up before you place it.
The Three Architectural Choices
Almost every Override team will pick one of these three patterns:
Architecture A — Sequential Single-Grip
Simplest
One universal manipulator handles both pins and cups, one at a time. Pick up a pin, drive to goal, place pin. Pick up a cup, drive to goal, stack on pin. Repeat.
Pros: Mechanically simple. One motor. Few moving parts. Fast to build. Easy to drive.
Cons: Two field traversals per scored stack (pin trip + cup trip). Slowest cycle time of the three.
Architecture B — Dual-Grip
Fastest
Two specialized grippers on the robot, each tuned to one object. Pick up a pin AND a cup before traversing the field. Place pin, then stack cup, all in one trip.
Pros: One field traversal per scored stack. Each gripper optimized for its object's geometry. Fastest cycle.
Cons: Two motors instead of one. More mechanical complexity. More to debug. Larger envelope.
Architecture C — Hybrid Intake-and-Grip
Compromise
A passive ground intake (rollers or compliant scoop) captures whichever object the robot drives over. An active grip extracts and places. Object identification done by sensor.
Pros: Faster pickup than Architecture A (no precise alignment needed). Single grip for placement.
Cons: Intake design is harder than it looks for two object shapes. Sensor logic adds programming complexity.
Sections 2 and 3 unpack each architecture with build details. Sections 4–6 cover the cross-cutting design problems (cup orientation, sensors, build sequence) that affect all three.
What This Deep Dive Covers
Game Elements — pins and cups in detail, with manipulation implications
Three Architectures — full analysis of A, B, and C with pros/cons and decision framework
Cup Orientation — the opaque/transparent problem and how to solve it
Sensors & Vision — AprilTags, optical sensors, and what each does for you
Build Order — what to prototype first, what to add later
Failures & Tuning — the five most common Override manipulator failure modes
STEM & Notebook — what concepts this teaches, plus interview lines and a quiz
1 of 8
// Section 02
Game Elements 🎯
What the pins and cups actually look like and what their geometry means for manipulation.
The Pin
A pin is a small two-color object that sits inside a goal. There are 63 pins in total, in four color combinations. Pins are the primary scoring object — visible red and blue pins each count for 5 points, and yellow pins are worth either 0 or +10 depending on the Toggle in their quadrant.
Pin Type
Count
Strategic Value
Red/yellow
20
Common. Red shows when placed correctly — 5 pts red. Yellow side gives toggle bonus potential.
Blue/yellow
20
Common. Same as above, blue alliance.
Yellow/yellow
19
Most common. Both sides yellow — only valuable when a Toggle claims the quadrant.
Red/blue
4
Extremely scarce. Both sides score for opposing alliances. Strategic resource — whichever side is showing scores 5 points for that alliance. Defensive denial of these pins is high-value.
From a manipulation standpoint, all pins share the same geometry. The color combination doesn't change how you grip them — just which pin you'd want to grip in any moment. Specific pin dimensions, mass, and exact shape are not in the reveal video transcript and will be confirmed in the manual.
The Cup
A cup is hollow and hourglass-shaped — narrow in the middle, flared at both ends. One side is opaque, the other is transparent. Cups stack on top of pins on goals. Adding a cup to a goal increases that goal's pin capacity (more cups = more pins can be stacked there).
The opaque/transparent rule is the one that drives mechanism design:
Pin in transparent side of cup → pin's color is visible → pin scores
Pin in opaque side of cup → pin's color is hidden → pin does not score
This means: if your robot places a cup with the opaque side toward the pin, you've just scored 0 points instead of 5. Cup orientation control is mandatory, not optional. See Section 4 (Cup Orientation) for how to solve this.
It also means a defensive play exists: if you can't prevent an opponent from placing a pin, you might be able to top it with a cup's opaque side to block the pin from scoring. Whether this is legal depends on the manual's descoring rules.
Real-world parallel: a cup is shaped like an old-fashioned hourglass timer or a cocktail shaker. The narrow middle is the "waist." The two ends are wider — one of them you can see through (like a clear plastic side) and the other is solid (like an opaque colored plastic side). When the cup sits on a pin, the pin pokes up into one of the ends — the question is which end.
The 1+1 Possession Limit
The reveal video states explicitly: at any given time, each robot is only allowed to possess one cup and one pin. This is the central design constraint of Override and shapes every mechanism choice.
What this means in practice:
No magazine accumulators. A common pattern from past games — pick up many objects, drive once, score them all — is illegal here. You can hold one of each, period.
Sequence matters. If you pick up a cup first and then can't reach a pin, you have to stack the cup on a pin that's already been placed (yours or someone else's).
Pin must come first in the stack. A stack is "pin in goal + cup on top." Order is enforced by the geometry.
Most efficient cycle holds both before the trip. Architecture B (dual-grip) is the only way to score a complete pin+cup stack with one field traversal.
The Goal
9 goals total in 3 sizes. 5 are neutral (either alliance can score in them) and 4 are alliance-specific (2 red, 2 blue). Every goal has an AprilTag for vision sensor identification.
No stacking limit. The reveal video says "there is no limit to the number of scoring elements that are allowed to be placed on a goal." You can keep stacking pin/cup pairs arbitrarily high.
Goal heights vary. 3 sizes confirmed; specific heights pending the manual. Your manipulator's reach must accommodate the tallest goal you plan to score on.
AprilTag mounting. Confirmed on all 9 goals regardless of alliance. See AprilTags Deep Dive for ID mapping (TBD post-manual).
What we still need to know: exact pin and cup dimensions, exact masses, exact goal heights for each of the 3 sizes, and field placement geometry. These come from the manual on Monday April 27. Until then, prototype with parametric CAD — build models with adjustable dimensions so you can update fast when the numbers ship.
2 of 8
// Section 03
Three Architectures 📍
Side-by-side analysis of the three Override manipulator patterns. Pick by your team's building skill, motor budget, and match strategy.
Architecture A — Sequential Single-Grip
One universal manipulator. Picks up either object. Two trips per scored stack: trip one for the pin, trip two for the cup.
What it looks like
A single gripper at the end of an arm or lift. The gripper has to be designed to hold both a pin and a cup — usually with a compliant inner surface (foam, rubber, or articulating fingers) that conforms to whatever shape it's clamping. Common implementations: a wide V-jaw with rubber pads, a passive scoop with a one-way trap, or a paired-finger gripper with adjustable spacing.
Cycle sequence
Drive to pin source (loader, field tile, or driver station)
Grip pin
Drive to goal
Place pin
Drive to cup source
Grip cup — orient if needed (see Section 4)
Drive back to same goal
Stack cup on pin
When to pick this
Rookie team or first-year V5 build — build skill is the limiter, simplicity wins
Your robot's strategy emphasizes other systems (drivetrain, defense, endgame) and the manipulator is "just good enough"
You want to ship a working V1 fast and iterate to dual-grip in Phase B
You're short on motor ports for the rest of the robot
Spartan's V1 Hero Bot will likely use Architecture A. The team's "roller-changing arm" concept fits this pattern — one mechanism that handles both objects sequentially. Phase B teams can graduate to B or C.
Architecture B — Dual-Grip
Two specialized grippers, each on its own actuator. Pick up one of each, traverse once, place both.
What it looks like
The robot has two distinct grippers. The pin gripper is small, narrow, optimized for the pin's geometry. The cup gripper is wider, often with a contoured inner surface that captures the cup's hourglass shape. The two grippers are typically stacked vertically or side-by-side on the same lift assembly — they share the lift but each has its own grip actuator.
Cycle sequence
Drive past pin source — pin gripper engages
Drive past cup source — cup gripper engages
Drive to goal
Place pin first
Stack cup — rotate first if orientation is wrong
When to pick this
Experienced team that has built multi-mechanism robots before
Cycle time matters most for your strategy (high-volume scoring)
You have motor ports to spare (typically 2 grip motors + 1 lift = 3 mechanism ports)
Drivers can handle a more complex robot with more state to track
Architecture C — Hybrid Intake-and-Grip
A passive ground intake (rollers, compliant scoop, or chute) captures whichever object the robot drives over. An active grip extracts the captured object from the intake and places it on the goal. Object type is identified by an optical or distance sensor.
What it looks like
The bottom of the robot has an angled scoop or roller intake near the chassis floor. The robot drives over a pin or cup; the intake captures it and feeds it into a holding zone on the chassis. Above the holding zone, an arm with a single grip extracts and places. A sensor on the holding zone identifies whether it's a pin or cup, and the robot programs accordingly.
Cycle sequence
Drive over pin source — intake captures pin
Identify (sensor confirms it's a pin)
Drive over cup source — intake captures cup
Identify (sensor confirms it's a cup)
Drive to goal
Extract pin from holding zone, place on goal
Extract cup, orient, stack on pin
When to pick this
You want fast field-pickup without precise alignment (common in chaotic match flow)
You have programming bandwidth for sensor logic and state management
You don't have the build skill for two precision grippers (Arch B) but want better cycle time than Arch A
You want a competitive Phase B robot that simplifies pickup at the cost of some placement complexity
Decision Matrix
Question
Arch A (Single)
Arch B (Dual)
Arch C (Hybrid)
How many motor ports for the manipulator?
1
2
2
Build complexity (1=easy, 5=hard)
2
4
3
Programming complexity (1=easy, 5=hard)
2
3
4
Average cycle time per pin+cup stack
Slowest
Fastest
Medium
Pickup precision required
High
High
Low
Driver workload
Lowest
Highest
Medium
Prototype risk if first attempt fails
Low — easy to fix
High — redesign costs time
Medium
For Spartan's 6-team program with mixed experience levels, the recommendation depends on team-by-team skill: rookies to Architecture A, experienced returners to B or C. Cross-team consistency is less important here than for the drivetrain — teams can have different manipulator architectures without breaking shared learning.
3 of 8
// Section 04
Cup Orientation 🔲
The opaque/transparent problem. Why it matters for scoring and how to solve it mechanically.
Why Cup Orientation Matters
The reveal video is explicit: "placing a pin into the opaque side of a cup prevents that color from being scored." This means a cup placed wrong-way-up on a pin is the same as not placing a cup at all (from a scoring standpoint). Worse, if you do this on every cup placement, you score zero points from your cup work for the entire match.
Your manipulator must do one of three things to handle this:
Always pick up cups in a known orientation — if cups always come out of the loader transparent-side-up, then your gripper just maintains that orientation through placement. Easiest solution.
Detect orientation, rotate if wrong — sensor identifies opaque vs transparent side, gripper rotates the cup before placement if it's wrong. Most robust but most complex.
Always rotate to a known state — gripper always rotates the cup 0 degrees or 180 degrees before placement to match a fixed reference. Simpler than detection but assumes you know the starting orientation.
Which option you pick depends on how cups are presented to your robot. We don't yet know the loader mechanics from the manual — if loaders deliver cups in a known orientation, Option 1 works. If cups can come from the field in any orientation (e.g., descored from another team's goal), Option 2 is necessary.
Detecting Cup Orientation
Two viable sensor approaches:
Optical sensor (V5 Optical 276-7405). Mount facing the cup's grip area. Detects color/brightness difference between opaque and transparent sides. The transparent side will read as "ambient" (whatever is behind the cup); the opaque side will read as the cup's solid color. Tunable threshold determines orientation. See Optical Sensor Deep Dive.
Distance sensor. Mount facing the cup's grip area. The transparent side passes some IR light through; the opaque side reflects more. Distance reading differs between the two sides. Less reliable than optical for this specific case.
Optical is the standard answer here. It's purpose-built for color/material discrimination and the opaque/transparent distinction is exactly the kind of contrast it was designed for.
Rotating the Cup
If your sensor detects the cup is wrong-way-up, your gripper needs to flip it. Three implementation patterns:
Wrist Motor (continuous rotation)
A small motor (5.5W or 11W with green cartridge) rotates the gripper itself around its wrist axis. Software commands a 180° rotation when the sensor reports wrong orientation.
Pros: Precise. Continuous rotation supports any starting angle. Can correct partial misalignments.
Cons: Adds a motor port. Wiring routes through a rotating joint (hard).
Pneumatic Wrist Flip
A double-acting pneumatic cylinder rotates the gripper between two fixed positions (0° and 180°). Single button or auto-trigger flips it.
Pros: Fast. No motor port consumed. Two discrete positions are usually enough for orientation.
Cons: Requires pneumatics infrastructure (compressor, tank, solenoid). Air budget must account for flip frequency.
Pre-Pickup Orientation
Don't flip after pickup — flip before. Sensor detects orientation while the cup is still on the field or in the loader; the robot approaches from the side that captures the cup correctly. No moving rotation mechanism.
Pros: Simplest mechanically. No rotation hardware. No motor port.
Cons: Requires drive precision to approach from the correct side. May not work if the cup is moving or in a constrained loader.
Defensive Cup Placement (If Legal)
The flip side of the orientation rule is that you can deliberately place a cup wrong-way-up over an opponent's pin to deny them the points. Whether this is legal depends on the manual's descoring and interaction rules.
If legal, this becomes a significant defensive play. Your robot would need a mode where it deliberately places cups in their opaque-up orientation. The implementation is the same hardware as orientation correction — just used inverted. Wait for the manual before designing strategy around this.
4 of 8
// Section 05
Sensors & Vision 🔮
AprilTags on every goal. Optical sensor for cup orientation. Distance sensor for goal alignment. What each does and how they cooperate.
AprilTag Identification on Goals
The reveal video confirms: "all of [the goals] are equipped with AprilTags for vision sensor identification, regardless of starting alliance." This is significant for an autonomous and skills program. The AI Vision Sensor (276-8659) reads up to 38 distinct AprilTag IDs (numbered 0-37) within a ~73° field of view.
Specific tag IDs and goal-to-ID mapping are not yet published — expected in the manual or via VEX's game-specific update. Once known, your manipulator's autonomous code can navigate to a specific goal by ID, identify it visually, and align before scoring.
For now, design the manipulator's programming to support AprilTag-based alignment without committing to specific IDs. See AprilTags Deep Dive for the integration patterns.
Optical Sensor for Cup Orientation
As discussed in Section 4, the V5 Optical Sensor (276-7405) is the right tool for opaque/transparent discrimination. Mount it inside the cup gripper, facing the inner surface of the captured cup.
Tuning:
Brightness threshold — transparent side reads through to ambient/whatever is behind; opaque side reflects bright cup color. Set threshold based on your specific cup color and ambient lighting.
Calibration ritual — before each match, run a calibration that captures the "clearly opaque" reading and the "clearly transparent" reading. The midpoint is your threshold.
Multi-sample averaging — take 5–10 readings and average. Single readings can flicker due to ambient light changes; averaged readings are stable.
Distance Sensor for Goal Alignment
For close-range goal alignment (the last 6″-12″ before placement), a distance sensor often beats vision. Mount one facing forward on the manipulator. Drive to within sensor range of the goal, then close the loop on distance:
Coarse approach: AprilTag-based heading and approximate distance via vision.
Fine approach: distance sensor confirms exact distance from manipulator face to goal rim.
Place: motor commands placement only when distance is in a tight tolerance (e.g., 2″ ± 0.5″).
This is the standard sensor stack for VEX manipulators that need both global navigation and local precision.
Color Sensor for Pin Identification
If your manipulator picks up pins one at a time and you want to identify which pin you're holding (for strategic decisions: e.g., is this a red/blue dual or a yellow/yellow?), a color sensor on the gripper interior reads the pin's color combination.
This is optional. Many teams skip it and let the driver visually verify before pickup. Depends on your cycle time and driver workload tolerance.
Sensor Stack Summary
For a competitive Override manipulator, the typical sensor configuration is:
Sensor
Role
Required?
AI Vision Sensor (AprilTag)
Goal identification & coarse alignment
Highly recommended — auton scoring depends on it
Optical Sensor
Cup orientation detection
Required if not using pre-pickup orientation
Distance Sensor
Final alignment to goal
Optional — nice-to-have
Color Sensor
Pin color identification
Optional — for strategic awareness
IMU
Heading & turning accuracy
Required for chassis (universal)
5 of 8
// Section 06
Build Order 🛠
What to prototype first, what to add later, and how to test as you build.
The temptation with manipulators is to build the whole thing at once. Don't. Override's manipulator has at least three independent design problems (pin grip, cup grip, orientation control) and stacking them all at once means you can't debug any of them. Build in stages, test at each stage.
1
CAD the pin and cup geometries first. Use parametric models with adjustable dimensions. Until the manual ships exact dimensions, use the reveal video's descriptions and your best guess. When the manual lands Monday, update the parameters — the rest of the model adjusts automatically.
2
Prototype a single grip on either object. Don't try to grip both yet. Pick whichever is mechanically easier (probably the pin) and build a working grip in isolation. Mount it on a stationary fixture, not the robot. Test grip force, release reliability, and self-centering behavior.
3
Mount the working grip on a basic arm or lift. Now test it on a moving robot. The gripper that worked on a fixture will fail in different ways during driver-control turns and accelerations — finding those failure modes early is the point.
4
Add the second grip. If you're going Architecture B, build the cup grip now. If you're going Architecture A, evolve the single grip to handle both objects. Test pickup & release on each.
5
Add cup orientation handling. Either pre-pickup approach logic (drive control), in-grip rotation (motor or pneumatic), or sensor-based detection. Test on a stationary cup; verify your robot consistently places it transparent-side-up.
6
Add scoring sensors. AprilTag for goal ID. Optical for cup orientation if not using pre-pickup logic. Distance for final alignment if needed. Test each sensor in isolation, then combined.
7
End-to-end cycle test. Pick up pin, drive, place. Pick up cup, drive, stack. Time the full cycle. If it's longer than 8 seconds, look for what's slow — that's your next iteration target.
8
Defensive robustness test. Have a teammate try to disrupt your robot during a cycle. Do they knock the pin or cup loose? Does the gripper survive contact? Defensive contact is part of every match — design for it.
6 of 8
// Section 07
Failures & Tuning ⚠
Five Override-specific manipulator failure modes and how to fix them.
Cup Drops During Travel
Robot picks up the cup successfully, but mid-drive (especially during turns), the cup slips out of the gripper. Score lost.
Fix: the cup's hourglass shape means it has limited points of friction contact with a flat-faced gripper. Add compliant material (foam, rubber) lining the gripper's inner faces. If the cup is still dropping, switch from grip-by-friction to grip-by-shape — wrap the gripper around the narrow waist of the hourglass.
Pin Goes In Wrong Side of Cup
Robot places pin and stacks cup successfully, but the pin's color is hidden because the cup is opaque-side-down. Zero points scored.
Fix: this is the cup orientation problem from Section 4. Either add an optical sensor with rotation correction, use pre-pickup orientation logic, or implement always-rotate-to-known-state mode. Without one of these, you score zero points from cups.
Stack Topples Mid-Score
Robot places pin successfully, but when stacking the cup, the previous stack topples or the new stack misaligns. Multiple cups/pins fall off the goal.
Fix: placement precision matters. Add a distance sensor for final alignment. Reduce drive speed during the final 6 inches of approach. Make sure the gripper releases at exactly the right height — releasing too high causes the cup to drop and topple.
Manipulator Won't Reach Tallest Goal
The robot can score on neutral and short goals, but the tallest goal is just out of reach. Strategy locked out from a major scoring zone.
Fix: CAD the manipulator's reach against confirmed goal heights (post-manual). If reach is short, options are: extend the lift, add a 2-stage cascade lift (see Cascade Lift Deep Dive), or switch to DR4B (see DR4B Deep Dive) for compact tall reach.
Robot Holds Two Cups (Or Two Pins)
Robot accidentally captures a second pin or cup while still holding one. Possession violation. Penalty.
Fix: the 1+1 limit is a rule, not a guideline. Your intake or grip mechanism must mechanically prevent capturing a second of either object while holding one. Options: gate the intake closed once an object is captured, position the second-grip out of contact range until the first slot is free, software lockout on intake motor based on holding-zone sensor.
Match-Day Tuning Sequence
Calibrate optical sensor against ambient lighting at the venue. Field lighting differs from your build space.
Verify AprilTag detection at the actual goal positions. Some venues have lighting that makes tag detection unreliable.
Test grip force with a fresh battery. Battery sag during practice can hide grip force issues that emerge mid-match.
Time a full cycle in match conditions. Cycle time longer than 8 seconds means you score 8–10 stacks per match instead of 12–15. Look for what's slow.
Practice cup orientation correction as a separate skill. Have a teammate hand you cups in random orientations; verify your robot always places transparent-side-up.
7 of 8
// Section 08
STEM & Notebook 📚
What this mechanism teaches, what to write in your engineering notebook, and how to talk about it with judges.
STEM Highlight — Engineering Tradeoffs
⚙ STEM Highlight
The Override manipulator decision is a clear example of engineering tradeoffs — the engineering principle that you usually can't get everything you want at once. Each architecture trades one thing against another:
Architecture A trades cycle speed (slower) for simplicity (easier to build). Architecture B trades simplicity (more complex) for cycle speed (faster). Architecture C trades both partially — faster pickup at the cost of programming complexity. There's no single "best" answer because the right choice depends on what your team values most.
Engineers face this constantly. A faster car uses more fuel. A lighter phone has less battery life. A cheaper material is less durable. Recognizing the tradeoff and choosing deliberately — rather than pretending one answer is always right — is what good engineering looks like. The decision matrix in Section 3 is what real engineers use to make these choices: list the things you care about, score each option, pick the one that wins on the things that matter most to you.
Interview Talking Points
How we picked our architecture: "Override has a 1-cup-and-1-pin possession limit, so we couldn't use a magazine accumulator. We had to choose between three patterns: a single grip that handles both objects (slower but simpler), two specialized grips (faster but more complex), or a hybrid intake design. We went with [your choice] because [your reasons]. The tradeoff was [what you gave up]."
The cup orientation problem: "The cup has an opaque side and a transparent side. If we place a pin into the opaque side, the pin's color doesn't score. So our manipulator has to control cup orientation. We use [optical sensor / pre-pickup approach / pneumatic flip] to make sure every cup we place has the transparent side toward the pin. Without this, we'd score zero points from our cups."
Why our choice fits our team: "We're a [first-year / experienced] team with [N] motor ports available after the drivetrain. Our match strategy emphasizes [cycle speed / defense / endgame]. Architecture [A/B/C] fits our skill level and strategy because [your reasoning]. If we were a different team, we might pick differently — that's the point of an engineering decision matrix."
Check for Understanding
Your team picks Architecture A (single grip) for V1 to keep things simple. The robot scores reliably but cycle time is 11 seconds per pin+cup stack. Average matches score 8 stacks. What's the most likely first upgrade for Phase B?
Switch to red cartridges for higher torque on the gripper motor
Add a second specialized grip (move toward Architecture B) so each cycle is one trip instead of two
Buy a faster motor to spin the existing grip faster
Add a third possession slot to hold more pins
Why: The bottleneck in Architecture A is the two field traversals per stack. Switching to dual-grip cuts that to one traversal — potentially halving cycle time. A faster motor doesn't help because the gripper isn't the slow part; the drivetrain traversal is. Higher torque doesn't help unless the gripper is failing to grip (it's not). And adding a third possession slot is illegal — the 1+1 rule is a hard rule.
What to Put in Your Engineering Notebook
The architecture decision matrix. Show the comparison from Section 3. Write down the team values you weighted. Show why the chosen architecture won.
Cup orientation sensor calibration data. Optical sensor readings for opaque side vs transparent side, your threshold, how it differed across lighting conditions. Numbers and graphs are notebook gold.
Cycle time evolution. Track cycle time as you iterate. The graph that shows 14 sec → 11 sec → 8 sec across iterations is exactly what judges want to see.
Failure log. Every time the manipulator failed during testing, note what happened, what you changed, and whether it worked. This is the "iteration" story judges look for.
The 1+1 possession limit constraint. Document how you ensure the robot doesn't accidentally exceed the limit. This shows judges you read the rules carefully.