System desgin project

ESP32 Autonomous Mobile Robot (AMR)

An advanced Autonomous Mobile Robot firmware built on the ESP-IDF v5.x framework. Designed with strict real-time constraints, thread-safety, and modular architecture, this project demonstrates professional-grade embedded systems engineering using FreeRTOS and Object-Oriented C++.

🚀 System Architecture

The firmware is designed with a strict Application-Library Separation.

• lib/: Contains hardware-agnostic, non-blocking drivers (Motor Control, PID, ADC DMA, Loadcell, Wi-Fi). No FreeRTOS tasks are spawned here.
• src/: The Application layer orchestrating high-level state machines, task creation, and global resource management.

Key Engineering Highlights

• Deterministic Real-Time Execution: Motion control and PID loops run precisely at 100Hz in a pinned Core 1 task.
• Thread Safety: Lock-free passing where possible, combined with strict portENTER_CRITICAL Spinlocks for sharing high-frequency data (like Encoder ticks and ADC raw values) across multicore tasks.
• Zero Dynamic Allocation in Critical Loops: No malloc or new in polling loops. Memory is strictly stack-allocated or statically provisioned at startup to prevent Heap Fragmentation.
• Direct Memory Access (DMA): The 5-channel Infrared Line Sensor array is sampled at 20kHz continuously via ESP32's ADC DMA hardware, entirely offloading the CPU.

🧠 Control Algorithms

1. Velocity Control (Low-Level)

• Incremental PID Controller: Controls the TB6612 H-Bridge drivers coupled with Quadrature Encoders.
• Slew Rate Limiting: Prevents sudden current spikes (Back-EMF) by artificially capping the maximum acceleration dv/dt.
• Integral Anti-Windup: Prevents overshooting when wheels are mechanically jammed or lose traction.

2. Autonomous Navigation (High-Level)

• Line Tracking: Uses Weighted Average (Centroid) calculation with Affine Calibration to determine precise offset error (e2).
• State Machine: Seamlessly transitions between MOVING_TO_PICKUP, WAITING_FOR_PACKAGE (Loadcell integration), and DELIVERING.
• Two-Phase Pivot Turn: Executes precise turning maneuvers at intersections without the need for IMU/Compass, relying purely on encoder odometry and side-sensor polling.

3. Remote Telemetry & Tuning

• Utilizes asynchronous UDP Sockets to broadcast telemetry data (RPM, Duty Cycle, Sensor Arrays) to a local network at high frequency.
• Supports Live-Tuning: Send a JSON payload via UDP to dynamically update PID gains, steering RPMs, or loadcell thresholds without recompiling or halting the robot.

🛠️ Build & Flash Instructions

This project is configured using standard ESP-IDF CMake.

1. Prerequisites: Ensure ESP-IDF v5.x is installed.
2. Set Target:

   idf.py set-target esp32

3. Build & Flash:

   idf.py build
   idf.py -p PORT flash monitor

📁 Repository Structure

├── MCU/
│   ├── include/           # Shared structures, Configs, and Orchestrator Headers
│   ├── src/               # FreeRTOS Tasks (Motion, Telemetry, UDP) & Main
│   ├── lib/               # Independent Component Libraries
│   │   ├── adc_dma/       # Continuous DMA ADC Driver
│   │   ├── loadcell_hx711/# HX711 Weight Sensor
│   │   ├── tb6612_encoder/# Motor & Quadrature Encoder Driver
│   │   ├── velocity_pid/  # Slew-Rate Limited PID
│   │   └── wifi_manager/  # Wi-Fi Station & AP provisioning
│   └── CMakeLists.txt     # Global IDF Component Registration
└── UI/                    # Node.js Server & Web Dashboard for UDP Telemetry