FLL Techniques

What scores on the competition table isn't clever algorithms - it's runs that do the same thing every time. This lesson combines everything so far into FLL-specific plays.

Reliability > speed > elegance

A robot game match is 2:30, scored by mission completion state (referees also observe the team's Core Values behavior). A mission that succeeds 9 of 10 attempts beats one that's 5 seconds faster but succeeds 6 of 10. Every technique here serves one word: repeatability.

Position calibration: fighting error accumulation

Every segment driven adds position error. Calibration = using a known landmark to zero it out.

Wall squaring

Back into the wall; the wall is straight, so a firm contact fixes both heading and position:

define wallSquare
set movement speed [30] %
start moving [backward]
wait [1.5] seconds                ← long enough to press in; wheel slip is fine here
stop moving
reset yaw angle [0]               ← pose is known: zero the gyro

Key points:

• Low speed (25-35%): a full-speed hit bounces and sheds parts.
• The robot's tail must be flat (two contact points touching together); a skewed approach jams at an angle.
• With a force sensor, replace the fixed wait with "wait until pressed" - faster and confirmed.

Line calibration

Drive perpendicularly across a known black line; the instant you see it, your position along the driving direction is known:

start moving [forward]
wait until <color sensor [C] reflected light < [blackThreshold]>
stop moving
move [forward] for [3] cm          ← fixed offset from line edge to line center, etc.

The two-sensor version also fixes heading: if one sensor hits the line first, the robot is skewed - stop the early wheel until the other catches up ("line squaring").

Launch and return

• Start jig: build a LEGO alignment frame that sits against the base wall; place the robot against the jig every launch. Placement error drops from centimeters to millimeters. Highest value-for-effort "technique" in FLL.
• Launch heading = gyro zero: square the robot, then press start.
• Return home fast: the trip home needs no pose precision - full speed, roughly aimed at base. Per the BIOGLOW rules, if you interrupt a run to relaunch, the robot and everything it carries must be completely inside Home, or you lose a Precision Token.

Structuring a run

A common workflow (not a rule): one launch (run) per program slot. The standard skeleton:

when program starts
init                        ← My Block: motor setup + threshold variables + reset yaw
gyroStraight [60]
turn [90]
follow [30]                 ← line follow the final approach for lateral precision
attachment action
wallSquare                  ← if a wall is on the way, wipe the error before the next model
move [backward] for [70] cm ← back to base (speed over pose precision here)

Principles:

• Pick the right navigation per segment: long haul = gyro straight, final approach = line/distance sensor, return = speed over precision.
• Put failure-prone missions last: an early failure can wreck the field for everything after it.
• Attachment swap time is run time: drill swaps to under 5 seconds.

Robustness details

Problem	Countermeasure
Battery level changes speed	Full charge at competition; end every move on encoders/sensors, never time
Venue lighting	Recalibrate color thresholds on site; shield the sensors
Wheel slip	Gentle acceleration at launch; keep tires clean (wipe before matches)
Table wall differences	Leave margin in wall-squaring wait times
Shaky hands	Start jig + fixed roles (same person places, same person presses)

Testing discipline

• Test every mission 10 times in a row and record the success rate. Below 8/10 = not done.
• After any program change, retest from robot placement, not just the changed segment.
• Keep a log: mission, success rate, average time, failure causes. Let the data decide, not gut feeling.

Exercises

1. Build a start jig, measure end-position scatter over 10 launches, compare with freehand placement.
2. Write one complete run: launch → gyro straight → turn → line-follow approach → attachment action → return. Run it 10 times and log the success rate.

This is about as far as Word Blocks goes. For faster loops, real functions and PID, continue to Python.