Main project in collaboration with John Deere for the undergrad course "Design of Advanced Embedded Systems", which delves mainly into Design and Analysis of Algorithms, Digital Signal Processing, Shared-Memory Architecture, and Communication Interfaces.
Farmers face several challenges, including labor shortages, the need for precision in planting and harvesting, and time-consuming manual operations. Traditional tractors require constant monitoring and human input, increasing operational costs and the likelihood of errors in large-scale farming. There is a growing demand for autonomous systems that can handle repetitive tasks with accuracy and reliability.
The proposed solution consists of a tractor prototype that features navigation via waypoints powered by technologies such as wireless communication, camera-based positioning and angle detection, inertial measurement generations and encoder distance calculations.
The system employs a NUCLEO-H745ZI-Q development board to control the tricycle-style vehicle platform with synchronized front steering and differential rear drive. Motion control is achieved through dual PWM channels: TIM13 handles servo-based steering with angle-to-pulse width mapping, while TIM14 manages ESC motor control with bidirectional speed control through pulse width modulation.
The internal navigation system fuses data from two sensor streams: a wheel-mounted encoder for distance measurement and an MPU6050 IMU (via the I2C4 peripheral) for orientation tracking. The encoder data is preprocessed by an Arduino Nano and transmitted via CAN to the main MCU (received by means of the FDCAN1 peripheral), where raw counts are converted to distance using a counts-per-revolution calibration.
Orientation tracking implements a signal processing pipeline:
- Raw IMU measurements undergo offset compensation and scaling.
- A Kalman filter processes each axis to reduce sensor noise.
- Gyroscope integration uses trapezoidal approximation for yaw calculation.
- Continuous angle normalization maintains orientation within 0-360 degrees.
The system employs a MATLAB-generated pure pursuit controller operating at LINEAR_VELOCITY with a LOOKAHEAD_DISTANCE parameter. Real-time state estimation combines encoder distance and IMU yaw to maintain accurate position tracking in Cartesian coordinates, enabling smooth trajectory following in a snake-like pattern.
Position data comes from the John Deere Global Positioning System, which tracks a vehicle-mounted marker. Coordinates are transmitted through an nRF24L01 wireless link (using the SPI5 peripheral) between a secondary board (NUCLEO-F103RB) and the vehicle. To accommodate the relatively slow GPS update rate, the system implements a time-based proportional control strategy:
The vehicle advances at a fixed SPEED while the controller calculates movement durations proportional to the distance from target. Between movements, the system pauses for STOP_TIME milliseconds to ensure stable position readings. Waypoint arrival triggers an audio alert through a PWM-driven buzzer using TIM1.
Both modes utilize USART3 for debugging and system monitoring, providing real-time state information at 115200 baud.
For detailed information on the electronics design, including the KiCad PCB layout, please refer to the dedicated electronics repository.
| Component | Quantity |
|---|---|
| Servo Steering Robot Kit | 1 |
| NUCLEO-H745ZI-Q | 1 |
| Arduino Nano | 1 |
| nRF24L01 2.4GHz Wireless Module | 1 |
| MPU6050 IMU | 1 |
| GM 25-370 12V DC Motor with Encoder | 1 |
| MCP2515 CAN Controller with SPI Interface | 1 |
| CJMCU-1051 TJA1051 CAN Transceiver | 1 |
| Tower Pro MG996R Servo Motor | 1 |
| RC 30A Brushed ESC (Electronic Speed Controller) | 1 |
| HW-508 Passive Buzzer Module | 1 |
| 330 Ohm Resistor | 1 |
| Li-Po Battery 3.7V 2000mAh | 2 |
| Li-Po 2S Battery 7.4V 1500mAh | 1 |
| LM2596 Step Down Regulator 25W 3A | 2 |