AI summary by TwinMind

The Fastest Maze-Solving Competition On Earth

Veritasium

25m 22s

Historical Origins & Evolution

  • 1952: Claude Shannon created Theseus, electronic mouse solving maze using telephone relay switches
    • Mouse was magnet on wheels following electromagnet controlled by relay switches
    • Used trial and error strategy, registered information in memory
    • Could navigate directly to goal without false turns after exploration
    • Called one of first examples of machine learning, inspired AI field
  • 1977: IEEE launched Amazing Micromouse Maze Contest after translation misunderstanding
    • Attracted over 6,000 entrants initially
    • Only 15 reached finals in 1979
    • Garnered enough interest for nationwide evening news broadcast
    • Competition spread globally

Competition Rules & Structure

  • Mice must be fully autonomous: no internet, GPS, remote control, or human assistance
  • Size constraints: maximum 25cm length/width, no height limit
  • Prohibited: climbing, flight, combustion/rocket propulsion
  • Maze: 3m x 3m square, corridors 18cm wide
  • 2009: Half-size category introduced (12.5cm mice, 9cm corridors)
  • Maze layout revealed only at competition start, no code changes allowed after
  • Time limits: 7-10 minutes total, maximum 5 runs from start to goal
  • Major competitions: All Japan, Taiwan, USA's APEC

Search Algorithms & Strategies

  • Early wall-following strategy defeated by moving goal away from edges, adding freestanding walls
  • Depth-first search: runs deep into maze, turns back only when necessary
    • Eventually reaches goal but may miss shortcuts
  • Breadth-first search: finds shortest path but requires excessive backtracking
  • Flood fill algorithm (most popular):
    • Creates optimistic map with no walls initially
    • Draws shortest path to goal, updates when hitting walls
    • Marks distance from every square to goal
    • Follows trail of decreasing numbers to zero
    • Uses return trip for additional maze exploration
    • Leaves irrelevant areas untouched

Physical & Technical Evolution

  • Early sensors: tall unwieldy arms replaced by compact infrared sensor arrays
  • Motors: stepper motors replaced by DC motors with encoders for more power/less weight
  • Gyroscopes added for orientation (like compass), enabled by mobile phone technology
  • All turning based on gyro feedback rather than wheel pulse counting
  • Dust management critical: competitors use tape to clean wheels between runs
    • Tiny friction changes ruin precision at high speeds

First Fosbury Flop: Diagonal Movement

  • Mighty Mouse 3 first implemented diagonal cuts through maze corners
  • Required chassis reduction to <11cm wide (5cm for half-size)
  • New algorithm needed: guides mouse with "blinders on" approach
  • Diagonal navigation more crash-prone than wall-following
  • Transformed jagged turning paths into narrow straight-aways
  • Led to U-turn motions replacing stop-and-pivot turns
  • Exponentially increased possible turn combinations

Second Fosbury Flop: Vacuum Fan Technology

  • Mocomo 08 first used vacuum fan system in competition
  • Built from quadcopter/drone parts
  • Generates downward force 5x mouse's weight (typical mouse ~130g)
  • Enables centripetal acceleration approaching 6 G's (F1 car levels)
  • Allows corner speeds up to 7 meters/second
  • Out-accelerates Tesla Roadster (briefly)
  • Solved speed control problem that limited early 2000s mice

Modern Performance & Innovation

  • 2017 All Japan: Red Comet won by taking longer but faster path
    • Chose 5.5m longer route with fewer turns
    • Won by 131 milliseconds over shorter-path competitors
  • Current mice can complete maze in ~6 seconds
  • Bronze/silver mice found shortest path in 7.4 seconds
  • Ongoing experimentation: 6-wheel, 8-wheel, omnidirectional movement, computer vision
  • Quarter-sized micromice in development (would sit on quarter coin)
  • Four-wheeled mice: first won All Japan in 1988, became norm 22 years later

Competition Philosophy & Future

  • Problem evolved from simple maze-solving to complex robotics challenge
  • Combines major engineering disciplines: robotics, programming, embedded systems
  • 1980s prediction of "problem solved" proven wrong by continuous innovation
  • Accessible to hobbyists - can be done in living room without laboratory
  • Competition bigger than ever nearly 50 years after inception

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome