// the find
Skythinker616/foc-wheel-legged-robot
Open source materials for a novel structured legged robot, including mechanical design, electronic design, algorithm simulation, and software development. | 一个新型结构的轮腿机器人开源资料，包含机械设计、电子设计、算法仿真、软件开发等材料
A complete wheel-legged balancing robot project from a single developer, covering everything from SolidWorks mechanical design and custom PCB schematics to MATLAB/Simscape control simulations and ESP32/STM32 firmware. It's a self-contained build guide for a ~$80 USD two-wheeled inverted pendulum robot with optional FPV camera. Aimed at hardware hobbyists who want a working reference design rather than a platform to build on.
The LQR control implementation is properly derived — the MATLAB scripts show the actual linearization and gain matrix calculation (sys_calc.m, lqr_k.m), not just magic numbers pasted into firmware. The custom FOC motor driver board (STM32 + CAN bus) is a meaningful design choice: six drivers communicating over CAN keeps the main ESP32 clean and gives you proper torque control rather than PWM speed control. Hardware files are exported in both Altium and EasyEDA/LCEDA formats, which is unusually considerate for a Chinese-origin project targeting makers who mostly use LCEDA. The plug-and-play FPV module design — completely removable without affecting robot function — is the right call for an optional subsystem.
Abandoned in November 2023 with no activity since, so any issues you hit are yours to solve. The ESP32 main.cpp is where all the control loop logic lives and it has no documentation — if the LQR gains need retuning for your specific motors or weight distribution, you're reverse-engineering someone else's undocumented state machine. No calibration procedure for the MPU6050 IMU is documented, which will bite you immediately when the robot drifts. The Android app is bare-minimum: hardcoded IP, no error recovery, and requires you to build it yourself from a 2021 Gradle setup that fights with modern Android Studio.