EV3 Robotics›Level 4 · WRO Prep›Lesson 45
Level 4 · Lesson 45 · EV3-L04-4560 minutes · Block 6 of 6 · Mat: WRO 2026 — Robot Rockstars
When something goes wrong, carry on — do not stop dead.
Everything you have built assumes things work. The line is there, the motor turns, the sensor reports. When one of those stops being true, most programs wait forever for a thing that is never going to happen.
The argument is arithmetic. A round is a fixed length. A robot stuck at twenty seconds keeps what it had already scored and gains nothing from the remaining time. A robot that abandons one leg and moves on can still score everything after it. Giving up on a mission is a scoring decision, and it is usually the right one.
You have done this twice already without naming it: the timeout on the synch flag in lesson 38, and the abandon plan in lesson 17. Today it becomes a habit.
Walk the route and mark the places a run realistically breaks.
Mark the abandon points: for each mission, where can the robot be left, mid-mission, and still get to the next one? That is the position each recovery has to reach.
One idea, applied three times: every loop that could wait forever gets a way out that is not success. That is a watchdog.
The lost line. The follower loop keeps running but the sensor never sees the line. Detect it with a timer — start one when the leg starts, and if the line has not been seen for, say, two seconds, stop following.
Use a timer, not a count of loop passes. A pass count depends on how fast the loop runs, which changes with what is in it — and after lesson 28 you know the loop speed is not a constant.
Then search: a short arc one way, then a wider arc back the other. If the line appears, carry on. If it does not, that leg is over.
The stalled motor. Commanded to move, not moving — jammed, or up against something. Detect it by reading the motor’s rotation, waiting a moment, and reading again. If it has not changed while you are telling it to turn, it is stalled.
Back off, retry once, then skip. Never retry forever. A jam that did not clear twice will not clear on the tenth attempt, and the round is going.
The sensor that stops reporting. A disconnected sensor, or a slave that has dropped off the chain, reads a constant value. That is the tell: a reading that has not changed at all while the robot has been driving is not a reading. Fall back to driving the leg on distance and heading alone.
Recovery must be bounded too. A search pattern with no limit is the same bug as the infinite wait, wearing a helpful face. Every recovery gets its own timeout, and its failure path is abandon this leg and go to the next mission — never try again.
Do the lost line first, because it is the most likely failure and the clearest to test.
Change the follow loop so it ends on any of three conditions:
Then set a variable saying which of the three ended it, because the code after the loop has to behave differently for each. A loop that exits without saying why leaves you guessing at exactly the moment you cannot afford to.
Wrap the search pattern in a My Block called Find Line, returning whether it succeeded. Then every follower in the program can use the same recovery, and you tune the search once.
Abandoning is not stopping. It is getting to a known place and continuing with the next mission, and it needs to be designed.
Step three is the one that gets skipped, and it is what makes abandoning worthwhile. A robot that gives up on a leg has been wandering — its heading and position are exactly what you should not trust. Carrying on without re-establishing them turns one lost mission into every mission after it.
Give the whole run a watchdog as well, not just each leg. If the round length is nearly up, skip straight to whatever scores at the end — returning to HOME, or a final action worth points. A robot still attempting mission two when the clock stops has thrown away whatever it could have collected on the way home.
A recovery you have not triggered is a recovery you do not have. Do all four, and watch what the robot does.
Two things to watch for. Recovery that is too eager — abandoning a leg that would have been fine — costs you points you had. Recovery that is too patient costs you the rest of the round. Tune the timeouts against your leg times from lesson 41, not against a feeling.
Run the full route while somebody from another team sabotages one thing at a moment of their choosing, without telling you which or when.
The robot should lose one mission and complete the rest. Score it, and compare against a clean run.
The gap between those two scores is what today is worth. Before this lesson that gap was the whole round, because the robot stopped. Now it should be one mission.
Then read the log. It should say which mission was abandoned, and that record should match what the saboteur actually did.
Today’s entry:
That is the last technique in the course. Lessons 46, 47 and 48 are a rehearsal, a scrimmage against another team, and the final tournament. Nothing new is taught from here — what you have is what you compete with.
This model drives, so its challenges are run on a mat. Mats differ between branches — check you are looking at the one in your room.
WRO 2026 RoboMission Elementary — Robot Rockstars · official WRO game mat, 2362 × 1143 mm
The challenges name these places rather than distances, so the same challenge works on any mat:
Switch mats above and every route below is redrawn on the mat you chose.
Work through the challenges in order — each is harder than the last. The mission comes after all three, and it is meant to make you plan before you build.
Walk the route and mark where a run realistically breaks — the plaza at E because the robot passes it constantly, the corner at D and C because a bad arrival at one means a bad arrival at the other, and anything with a mechanism. Then mark the ABANDON POINTS: for each mission, where the robot can be left mid-mission and still reach the next one.
Give the follow loop three ways out — success, timeout, and line lost — with a variable saying WHICH ended it, because the code after the loop has to behave differently for each. Use a timer rather than a count of loop passes: a pass count depends on loop speed, which is not a constant. Then wrap the search pattern in a Find Line My Block that reports whether it succeeded.
Trigger every recovery on purpose, because one you have not triggered is one you do not have: cover the line mid-leg with paper, hold a mechanism so it cannot move, unplug a sensor, and unplug the chain cable. Recovery that is too eager throws away points you had; too patient costs the rest of the round. Tune the timeouts against your leg times, not against a feeling.
Run the full route while somebody sabotages one thing without telling you which or when. The robot should lose ONE mission and complete the rest — and the gap between that score and a clean run is what this lesson is worth, because before it that gap was the whole round. Abandoning is not stopping: log it, get somewhere known, RE-ESTABLISH position, then start the next mission. Skipping that third step turns one lost mission into every mission after it.