Design and Implementation of a Two-Wheel Self-Balancing Robot Using PID Control on an ESP32
Custom PCB, Simulink-generated controller, and a Wi-Fi control panel with OTA updates
DOI:
https://doi.org/10.31224/8857Keywords:
self-balancing robot, inverted pendulum, PID control, ESP32, Simulink Embedded Coder, FreeRTOS, over-the-air update, system identificationAbstract
This report describes the design and construction of a two-wheeled self-balancing robot, from the hardware to the control software. The robot runs on an ESP32 and reads its tilt angle from an MPU-92/65 IMU. It is powered by Li-ion cells, with an ATtiny13A as a secondary microcontroller that keeps the board in a safe state and resets the ESP32 during firmware updates. I designed a two-layer PCB in Altium Designer that also leaves room for a camera and a servo for later additions. The balancing controller is a positional PID built from basic Simulink blocks, with a low-pass filter on the roll signal. C code generated with Embedded Coder was integrated into a C++ program that uses FreeRTOS and ESPAsyncWebServer. A web control panel served by the robot lets the user change controller parameters, monitor the IMU data at about 200 Hz, read non-volatile system logs, and upload new firmware over the air. During testing, the geared DC motors stalled when the PWM command changed by more than 470 counts at once, so a rate-limiting stage was added before the motor output. Two physical models of the robot were tried, but neither gave usable PID gains. A transfer function identified from 120 s of roll data sampled at 200 Hz reached only a 24.49% fit. The gains were therefore tuned experimentally through the control panel, which gave a steady-state error of about 0.3 degrees. Remote driving and live camera streaming are planned as future work.
Downloads
Downloads
Posted
License
Copyright (c) 2026 Mohammad Sina Majidi

This work is licensed under a Creative Commons Attribution 4.0 International License.