Skip to content

backprop-pray/Tracky

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

120 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

🌱 Tracky β€” Autonomous AI Farming Rover

tracky final image

πŸ₯‡ 1st Place @ Hack TUES & TUES Fest
Built by Backprop & Pray

W W W


🚜 What is Tracky?

Tracky is a fully autonomous AI-powered rover that navigates farmland, detects plant diseases, and maps problem areas β€” helping farmers save resources, reduce waste, and increase yield.

It combines:

  • πŸ€– Reinforcement Learning navigation
  • πŸ‘οΈ Computer Vision (anomaly detection)
  • 🧠 Knowledge Graph + RAG system
  • πŸ“± Mobile app with real-time insights

🌍 The Problem

  • 🌱 ~1 billion tons of crops are lost globally due to preventable plant diseases

  • 🚜 Existing solutions are:

    • Expensive
    • Not fully autonomous
    • Hard to scale for small farms
  • πŸ‘¨β€πŸŒΎ 75–85% of farms worldwide are small-scale

    • Limited access to advanced technology
    • High cost of machinery
    • Inefficient use of pesticides

πŸ’‘ Our Solution

Tracky enables:

  • πŸ” Early disease detection
  • 🎯 Targeted treatment (decimeter precision)
  • πŸ’° Lower operational costs
  • 🌿 Reduced chemical waste

It’s designed to be:

  • Affordable (prototype < €50 excluding compute)
  • Compact & terrain-adaptive
  • Fully autonomous

🧠 System Overview

embedded image

πŸ”© Embedded System

  • Raspberry Pi-based control unit
  • Distance sensors for navigation
  • GPS module for precise mapping
  • High-resolution camera for plant monitoring
  • Custom 3D-printed chassis (assembled in 1 day)

πŸ€– AI Navigation (Reinforcement Learning)

  • Custom RL agent controls the rover
  • Learns terrain dynamically
  • Fully autonomous movement between crop rows

cool imave v2 as png


πŸ‘οΈ Vision System (Anomaly Detection)

good video

  • Learns farm-specific β€œhealthy baseline”
  • Detects deviations β†’ potential diseases
  • Continuously improves with new data

Output:

  • πŸ“ GPS-tagged anomaly
  • πŸ“Š Confidence + classification

🧠 Knowledge System (RAG + Graph)

graph pictograms
  • Connects:

    • Plant conditions
    • Historical cases
    • Treatments
  • Enables:

    • Smart recommendations
    • Continuous learning across farms

πŸ“± Mobile App

mobile app image
  • Live map of detected issues
  • Field analytics
  • Feedback loop β†’ improves models

πŸ”„ Self-Improving System

All components work together:

  • RL agent improves navigation
  • Vision model improves detection
  • Knowledge system improves recommendations

➑️ With every farm, every user, every day β€” Tracky gets better


πŸ’Ό Business Model

Tracky is not just a robot.

It’s a data + intelligence platform:

  • πŸ’Έ Low-cost rover (easy adoption)
  • πŸ“Š Subscription for:
    • Monitoring
    • Analytics
    • Disease prediction
  • 🌍 Data platform for researchers & bio labs

πŸ‘¨β€πŸŒΎ Built by Backprop & Pray

us the farmers

We are a team of students from TUES, building real-world AI systems from scratch.

About

πŸ† 1st place at HACKTUES 12 πŸ†: Tracky is a fully autonomous AI rover that detects plant diseases and other crop issues, helping farmers save time and resources. It also builds a global knowledge graph to support agricultural insights for the broader good. Build all in 48 hours

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors