Top Class

Maze league · Republic of Korea · RoboCup 2026

Document register

  1. Poster1 pagePublished
  2. Presentation videoYouTubePublished
  3. Bill of materials75 KBPublished
  4. Team description paperNot shared
  5. Engineering journalNot shared
  6. Source code140 KB · GitHubPublished

Sharing each document is the team's decision. “Not shared” means this team chose not to publish it, or did not submit one — not that it is missing from the archive.

Top Class's robot
The Top Class team

Poster

Read the text of this document — 622 words
RoboCupJunior Rescue Maze 2026 — Incheon

TEAM TOP CLASS
Autonomous maze rescue robot                  |   League: Rescue Maze                  |    South Korea
M5Stack Core S3 Lite • BaseX + 4x LEGO EV3 motors • 4x ToF • 2x Color • 2x UnitV2 AI cameras • servo rescue-kit dropper

  THE TEAM                                                                                                                 ROBOT CAPABILITIES
                                                                                                                           • Fully autonomous maze exploration with complete-coverage mapping
                                                                [USEONG KIM, HAYOUNG OH]                                   • Precise cell-by-cell navigation with gyro-PD straight driving & 90° turns
                                                                                                                           • Detects letter victims (Φ/Ψ/Ω) and cognitive targets (colour)
                                                                     - Hardware & mechanics
                                                                                                                           • Rescue-kit deployment via servo dropper
                                                                                                                           • Speed-bump traversal via differential suspension
                                                                [SOO LEE, HAYOUNG OH]                                      • Black-tile avoidance, checkpoint recovery after lack-of-progress resets
                                                                     - Software (mapping & motion control)

                                                                [SOO LEE, USEONG KIM]
                                                                                                                           SOFTWARE — MAIN LOOP
                                                                     - Vision & sensors (cameras, color, ToF)

                                                                Awards / results:                                                                                          START / restore checkpoint
                                                                     3rd place in Rescue Maze at the Korean competition                                                                    v

                                                                                                                                                                      Pick next cell (DFS: open/closed lists)
                                                                                                                                                                                           v

  HARDWARE — ROBOT & SENSORS                                                                                                                                             Turn + drive one cell (gyro-PID)
                                                                                                                                                                                           v

                                                                                                                                                                Read walls (ToF x4) + floor (Color x2) -> 4-bit map
                                                                                                                                                                                           v

                                                                      • Controller: M5Stack Core S3 Lite                                                              Scan side cameras (UnitV2) -> victim?
                                                                        (ESP32-S3, dual-core 240MHz, 9-axis IMU)                                                                           v

                                                                                                                                                                        If victim: deploy rescue kit (servo)
                                                                                                                                                                                           v
                                                                      • I2C hub: PaHub (PCA9548A) —
                                                                        splits the six same-address (0x29) ToF &
                                                                                                                                                                  Unexplored left? yes: loop     no: A* path home
                                                                        Color sensors

                                                                                                                           SOFTWARE — ALGORITHMS & TESTS
                                                                                                                           • Map: 2D grid; each cell is a 4-bit wall code (N/S/W/E); grows from the start cell (auto-expanding)
                                                                                                                           • Explore: DFS with open/closed lists; A* (heapq, Manhattan h) for the path home
                                                                                                                           • Motion: our own BaseX DriveBase (4 motors); gyro-PID straight driving & 90° turns
                                                                                                                           • IMU: BMI270 gyro — integration + 1° deadzone + auto bias re-zero
     • Drive: BaseX (STM32, I2C) + 4x LEGO EV3 motors,                                                                     • Robust: checkpoint save/restore after a reset; bidirectional drive avoids 180° turns
       12T→28T gearing, 4WD                                                                                                • Runtime: MicroPython, threaded I2C bus + uasyncio; 2x UnitV2 cameras for victim ID

                                                                                                                            How we test:
     • Vision: 2x UnitV2 edge-AI cameras (left/right)                                                                       • Desktop A* grid visualiser (astar/visualize.py) before hardware
       — victim & target                                                                                                    • On-maze tuning: cell distance, gyro 90° turns, ToF wall thresholds
                                                                                                                            • [TEAM: add measured results — laps completed, victims found, reliability %]

                                                                                                                           SPECIFICATIONS

                                                                                                                           Controller              M5Stack Core S3 Lite (ESP32-S3)

                                                                                                                           Language                MicroPython + Python (sim)
                                                                      • Distance: 4x ToF VL53L0X (front/back/left/right)
                                                                                                                           Sensors                 4x ToF, 2x Color, 2x UnitV2, 9-axis IMU
                                                                        — wall sensing
                                                                                                                           Construction            LEGO Technic + 3D-printed PLA (Fusion 360)

                                                                                                                           Development time        Early Apr 2026 - Jul 2026 (~3 months)
                                                                      • Actuator: 2x C017CLS micro-servo
                                                                        — rescue-kit dropper                               Material cost           [from BOM total — TEAM TO FILL]

                                                                                                                           DEVELOPMENT & BUILD PROCESS

                                                                                                                              • Designed custom parts in Fusion 360, printed in PLA (sensor mounts, dropper, brackets)

     • Floor: 2x Color TCS3472 —
       black-tile / checkpoint detection

                                                                                                                               • Built drivetrain on a LEGO Technic chassis with differential suspension

  HARDWARE HIGHLIGHTS
                                                                                                                              • Wrote our own BaseX motor driver from its I2C register map
  Innovative mechanisms (Rescue Maze):

                                              CUSTOM DRIVE — BaseX + EV3
                                              No ready-made MicroPython driver exists for the BaseX motor base,
                                              so we wrote our own from its I2C register map (basex.py: straight, drive,
                                              encoder odometry).
                                              Gyro-PD heading control is layered on top.

                                                                                                                              • Verified the maze logic on desktop (grid visualiser) and a motors-disabled dry-run before hardware

                                              DIFFERENTIAL SUSPENSION
                                              A central differential lets the left and right sides flex independently,
                                              keeping all four wheels on the ground over speed bumps and
                                              uneven tiles — reliable traction and odometry

                                              SERVO RESCUE-KIT DROPPER
                                              Two micro-servos deploy rescue kits to the left / right wall
                                              when a victim is detected.

Team Top Class • RoboCupJunior Rescue Maze 2026 • South Korea                                                                                                                                  DRAFT — replace [bracketed] text & dashed photo boxes before submitting

Open as plain text

1 page, rendered as images so they load quickly. The text above is the document's own, extracted from the PDF.

Download the original PDF (649 KB) from GitHub

Presentation video

Hosted on YouTube. The player loads only when you press play.

Open in YouTube

Bill of materials

Shown as the original PDF, because this one is smaller that way and its text stays selectable and searchable.

Open Bill of materials

Source code

The team's own source code, 140 KB. It is a download rather than part of this page, because a zip is something you open on your computer. It comes from GitHub, which some school networks block.

Download Top Class's source code