PRIMEX FCoEE
Document register
- Poster1 pagePublished
- Presentation videoYouTubePublished
- Bill of materials65 KBPublished
- Team description paper622 KBPublished
- Engineering journalNot shared
- Source code7.7 MB · 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.
In their words
PRIMEX FCoEE is a fully autonomous maze-solving rescue robot developed by a student team from the Faizuddin Center of Educational Excellence (FCoEE) for the RoboCupJunior Rescue Maze 2026 competition. The robot is built on a dual-processor architecture, with a Jetson Orin Nano handling all vision and navigation decisions and an ESP32-S3 managing real-time motor and actuator control, mounted on a fully custom chassis 3D-printed using engineering-grade carbon fiber reinforced materials (ABS-CF and PETG-CF) for structural rigidity and durability under field conditions. The robot explores unknown mazes autonomously using a Right-Hand Rule algorithm with real-time 2D grid mapping, requiring no pre-mapping and no hardcoded field logic. It detects two types of victims. The first victim is Greek letter victims (Φ, Ψ, Ω) using a custom TensorFlow deep learning model trained on a manually captured dataset covering varied angles, distances, and lighting, achieving 90% accuracy. The second victim is cognitive targets, detected using outermost-circle-first template strategy that anchors inner ring boundaries and computes the health status sum from color values, achieving 80% ring layer detection and 90% per-ring color classification. Floor tile types are reliably identified by a downward-facing mBot2 color sensor with local calibration performed before each run, allowing the robot to adapt to changing venue lighting conditions.
The full system integrates high-performance hardware, autonomous maze navigation, trained letter victim detection, template-based cognitive target detection, and adaptive floor color sensing into a single cohesive whole. The combination reflects the team's commitment to building a robot that is not only technically capable across every scoring dimension of the competition, but also robust and consistent under real field conditions.
Poster
Read the text of this document — 321 words
AFIF (LEADER) HAZIEQ MUQRI (MENTOR 1)
HARDWARE AND MOVEMENT LEAD GREEK LETTER DETECTION CONSULT–DESIGN & HARDWARE
HAFIY(CO-LEADER) ADAM AMEERUL (MENTOR 2)
MAPPING AND NAVIGATION LEAD COGNITIVE TARGET DETECTION CONSULT–ALGORITHM & STRATEGY
1 HARDWARE PART 200mm COMPONENT
CONFIGURATION
CF-REINFORCE MATERIAL FOR HORIZONTAL HARDWARE INNOVATION DUAL CAMERA SWAPPABLE POWER
CHASSIS SPRING-LOADED ADJUSTABLE ANGLE TOOL BATTERY DOCK
ABS- CF for body panels (high RESCUE KIT DEPLOYMENT
Two IMX219 160° wide- angle
stiffness, heat resistance), PETG- 20V 1.5Ah power tool battery
Spring- loaded magazine with dual cameras mounted left and right at
CF for chassis frame (toughness on a quick- release dock.
slide ramps (left and right). 100–110mm height with adjustable
under drive loads), PLA for Swappable in seconds
MG996R servo actuates the ejector 15–20° tilt. Enables continuous
accessories. betweenrounds withouttools
arm to bilateral victim wall scanning while
CF materials ensuring full
deploy kit moving — no stop-
reduce flex power capacity
within and- rotate needed. Tilt
under field for every
15cm of mechanism allows
impact. the victim. scoring run.
field- day angle tuning.
2 SOFTWARE PART SYSTEM ARCHITECTURE GREEK ALGORITHMS & METHODS COGNITIVE
LETTER TARGET
DETECTION DETECTION
S1 – Capture Dataset for 10,500 S1 – Detect outermost circle
images, 3500 for each letter S2 – Template matching into
class. 1:2:3:4:5 diameter ratio
S2 – Train TensorFlow CNN to S3 – NumPy masking per ring
generate model. with HSV dominant color.
S3 – Detectionbased on S4 – Sum weightage for all 5
trained model rings
The robot calibrates floor colors, initializes the 2D grid map, then navigates AUTONOMOUS MAZE NAVIGATION
autonomouslyusing the Right- Hand Rule. It responds to floor tiles in real time,
stops on blue, reverses on black, and saves checkpoints on silver. Both cameras RIGHTHANDRULE – EMPTY TILE– Recover
scan for victims simultaneously,deployingrescue kits and blinking the RGB LED Follow wall for exploring based on non- discovered
on detection. and mapping tile. tile from mapping
Faizuddin Centre of Educational Excellence RoboCup Junior 2026
1 page, rendered as images so they load quickly. The text above is the document's own, extracted from the PDF.
Presentation video
Hosted on YouTube. The player loads only when you press play.
Bill of materials
Shown as the original PDF, because this one is smaller that way and its text stays selectable and searchable.
Team description paper
Shown as the original PDF, because this one is smaller that way and its text stays selectable and searchable.
Source code
The team's own source code, 7.7 MB. 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.