e‑Yantra CropDrop Bot Prototype

Autonomous agro‑village delivery bot using Q‑Learning + PID control

Mission

Build an autonomous delivery bot for the “e‑Yantra Agro‑Village”. The bot navigates a dual-track system: Farm (white line) and Village (black line), collects crop crates using an electromagnet, identifies crates using a color sensor, and delivers them to correct aggregation centers while maximizing efficiency (“freshness points”).

Core Highlights

Reinforcement Learning (Q‑Learning): trained an agent on a grid map to learn optimal delivery routes.
PID Control: tuned PID to maintain stable line following in real hardware.
Dual-track Navigation: logic to switch between farm and village tracking.
Crate Interaction: color sensor-based crate identification + precise pickup/drop with electromagnet.

Task Breakdown (Task 0 → Task 6)

Task 0: Setup (STM32CubeIDE, Python, Git).
Task 1: PID tuning + Q‑Learning training in simulation.
Task 2: Convert Q‑tables to motion commands.
Task 3: Hardware testing: STM32 Nucleo F103RB, IR sensors, motors, color sensor, electromagnet, PCB.
Task 4: White-line & black-line navigation + transition logic.
Task 5: Crate identification + precision stopping + electromagnet pickup/drop.
Task 6: Final autonomous run optimizing time/efficiency (“freshness points”).

Technical Stack

Controller + Firmware

STM32 Nucleo F103RB
STM32CubeIDE
Embedded C/C++
PWM motor control + IO mapping

AI + Control + Hardware

Q‑Learning (RL)
PID controller tuning
IR sensors (line tracking)
Color sensor (crate ID)
Electromagnet (pickup/drop)

Certificate

This project contains a certificate. Click below to view/download it.

Open Certificate PDF Contact Me

Key Learnings

Simulation → hardware transfer
Q‑table route optimization
PID stability improvements
Sensor calibration in real arena
Reliable pickup/drop workflow

Interested in similar projects?

Start a Project