Contest strategy¶
Micromouse contests differ in the details that matter most. This page lists the rule questions that change the design, and the strategy the robot is built around.
Rules first¶
Everything in config.h assumes classic micromouse rules (16 x 16 cells of 180 mm, 12 mm walls, 2 x 2 goal in the centre, start in a corner with the exit to the north). Before any tuning, get the rulebook and answer these, because each one changes the design:
| Rule question | What it changes |
|---|---|
| Maze size and cell size | MAZE_W, MAZE_H, CELL, robot width limit, whether pivots fit; rerun the turn designer |
| Goal location and size | GOALS in main.cpp |
| Is the maze known before the run? | If yes, skip search: load the map and go straight to the fast run |
| Number of runs and total time limit | how many search round trips we can afford (trips < 4 in runSearch) |
| Scoring: best run, sum of runs, penalty for search time? | whether to search slowly and safely or to search fast |
| Touching / restarting penalty | how aggressive the fast-run speed can be |
| Wall material and colour, floor material | ToF calibration, tyre choice, max acceleration |
| Is wireless allowed during runs? | debugging telemetry over Wi-Fi/BLE |
| Size and weight limits | chassis |
How runs are won¶
- Finish every run. A DNF scores nothing. Pick the fast-run speed from ten clean runs out of ten, then go one step slower at the contest (
docs/BRINGUP.mdstep 7). - Search once, carefully. Search speed barely matters if scoring is the best fast run. The map is saved in flash after every leg, so a crash does not lose it.
- Fast run with smooth turns. Pivots stop the robot at every corner; smooth turns roughly halve the fast run in simulation. Diagonals are faster still but need a calibrated gyro and are the least reliable mode in simulation.
- Spend practice time calibrating, not coding. ToF offsets on the real walls, wheel radius on the real floor, gyro scale.
Preparation timeline¶
| Weeks before | Goal |
|---|---|
| 10 | Parts ordered (with spares) |
| 8 | Chassis built, bring-up steps 1 to 3 pass |
| 6 | Calibration and closed-loop motion tests pass (steps 4, 5) |
| 5 | Practice maze built (at least 6 x 6 cells with real dimensions) |
| 4 | Search works end to end on the practice maze |
| 3 | Fast run, ten clean runs at a conservative speed |
| 2 | Speed tuning, full spare set assembled |
| 1 | Freeze code. Only parameter changes. Rehearse the contest-day checklist |
If the contest is less than 6 weeks away, stay with smooth turns at the safe or normal preset (or pivots if the smooth runs are not clean yet), skip diagonals entirely and focus only on reliability.
Run log¶
Keep a spreadsheet: date, battery voltage, mode, speed setting, result (clean / touched / crashed), time, notes. Decisions about speed come from this log, not from memory.