A Model Predictive Control Rover

Post by Souleymane Dembele on Dec 01, 2024

An autonomous rover that avoids obstacles with real-time Model Predictive Control across two microcontrollers.

What it does

The rover senses with sonar and an MPU6050 IMU and avoids obstacles using a real-time MPC formulation. An STM32 (custom RTOS) handles motor control over PWM, while an ESP32 (FreeRTOS) runs the Hildreth constrained-optimization solver and a Wi-Fi telemetry server for live data. It cut obstacle-avoidance response time by about 30% over the earlier PID-based rover, and was validated in MATLAB simulation and on real hardware.

Links

This repository is private. Request access via LinkedIn.