Self-Balancing Bipedal Robot Firmware & Control Suite

This repository contains the complete firmware, hardware documentation, digital twin utilities, and graphical tuning suites for a Self-Balancing Bipedal Robot. The system combines an STM32-based 100 Hz embedded PID control loop with articulated Dynamixel AX-12+ legs, complementary IMU posture sensing, 3DR wireless telemetry radio, FlySky FS-iA10B RC remote control, and a Python GUI / Jupyter Digital Twin interface.

---

📸 Gallery & Demo

Meet PAW — the physical build of this project, 3D-printed and assembled on the bench:

Left: front view showing the two articulated Dynamixel-driven legs, wheel motors, and the 3D-printed chassis with carry handle. Right: top-down view of the chassis, showing the parallel-link leg mechanism and the orange gripper arms mounted at the front.

🎥 Demo video: a short clip of PAW balancing and moving under its own control loop. ▶️ Watch on Google Drive (streams inline — recommended), or download the raw file straight from the repo.

---

🤖 About the Robot

The Self-Balancing Bipedal Robot is a dynamic robotics platform balancing on two wheels attached to articulated, servo-driven legs:

• Dynamic Posture Adjustment: Alter height, stride separation, and lateral lean dynamically while preserving upright balance via inverted pendulum control.
• 100 Hz Embedded Control Loop: An onboard MPU6050 IMU calculates body pitch inclination, fused via a complementary filter ($\alpha = 0.96$). The STM32 drives L298N H-Bridge PWM outputs to wheel motors with quadrature encoder velocity feedback.
• FlySky iBUS RC & 3DR Wireless Telemetry: Real-time remote operation via FlySky FS-iA10B receiver over iBUS (Serial1), alongside non-blocking telemetry streaming over 3DR Radio (Serial3 @ 115,200 baud) using a pipe (|) terminator protocol.

---

⚡ Electronics & Hardware Architecture

• Microcontroller: STM32F103C8T6 (Bluepill) operating as the real-time core.
• Leg Actuators: 4× Dynamixel AX-12+ Smart Servos (IDs: 6, 0, 14, 1) communicating over a 1 Mbaud half-duplex UART bus (Serial2).
• Wheel Actuators: 2× 12V DC Motors equipped with Hall-effect Quadrature Encoders for wheel odometry.
• Motor Drivers: L298N H-Bridge PWM motor driver module.
• Sensors & Peripherals: MPU6050 I2C IMU (400 kHz), 3DR 433MHz/915MHz Telemetry Radio (Serial3), and FlySky FS-iA10B Receiver (Serial1 iBUS).
• Power Supply: 11.1V – 12V 3S LiPo battery for motors and servos, step-down BEC for 5V/3.3V logic.

(For detailed pin assignments, wiring schematics, and USART assignments, see docs/Hardware_Connections.md).

---

📁 Repository Organization & Core Submodules

The robot's working code lives at the repo root, in robot_control_suite/. Everything else — earlier rework iterations, the original baseline firmware, and one-off tuning/diagnostic tools that have been superseded — is archived under legacy/, kept for reference but not actively maintained.

1) robot_control_suite/ — main, active codebase

Unified leg control and balance tuning suite. Each subfolder is a self-contained firmware + GUI pair with its own serial protocol — see robot_control_suite/README.md before assuming anything crosses over between variants. RC-capable variants (drive the robot with a FlySky transmitter) sit directly here:

• rc_mcu_ik_wireless: Flagship — MCU IK + 3DR wireless telemetry + FlySky FS-iA10B RC receiver.
• rc_balance_fusion_wireless: mcu_balance_fusion_wireless with a FlySky transmitter added.

wireless_no_rc/ — wireless variants with no RC transmitter, GUI/bench-tuned only:

• mcu_ik_engine_wireless: Wireless 3DR GUI tuning module.
• mcu_ik_engine_pretest_wireless: Minimal single-loop wireless balancer; retains a PING/PONG latency probe (latency_test.py).
• mcu_balance_fusion_wireless: Sensor-fusion balancer (Stage 2, in progress). Grouped here on its own name/README; its main.cpp does contain an iBUS decoder (it's the parent rc_balance_fusion_wireless was forked from).
• mcu_pos_wireless: Position/encoder-target tuner.

Also present: blue_pill_dev/, black_pill_dev/ (staged bring-up sub-projects), ax12_control/ (standalone servo tool) — see each folder's own README.md.

2) legacy/ — archived, not actively maintained

Split into two subfolders by what the code is, not just when it was written:

• legacy/old_firmware/ — superseded full firmware/controller generations:
  • mcu_ik_engine, mcu_ik_engine_wired, pc_ik_engine: earlier robot_control_suite variants (on-host/wired IK), superseded by the wireless variants above.
  • Balance_Rework_firmware/ (+ Balance_Rework_README.md): earlier single-cascade rework firmware (colon-delimited telemetry), predates robot_control_suite.
  • PlatformIO_Firmware/: original baseline C++ firmware — motor control snippets, hardware tests, early PID implementations.
  • Python_Controller_Digital_Twin/: original Python Jupyter notebooks (bipedal_digital_twin_controller.ipynb), legacy AX-12 controllers (ax12_controller_legacy.ipynb), and initial communication scripts.
• legacy/testing/ — one-off tuning/diagnostic tools and test rigs, not full robot programs:
  • autotuner/: safety-aware PID tuning tool with automated disturbance testing and validation scoring.
  • mpu_inspector/: diagnostic GUI (mpu_inspector_gui.py) and Web Serial interface (mpu_inspector_web.html).
  • servo_home/: servo-homing/calibration test rig (leg_control.py, digital_twin_legs.py).
  • temp_stm32_uploads/: scratch STM32 upload experiments.
  • latency_log.csv: recorded output from a past latency test run.

3) docs/ — reference documentation

• Robot_Specification.md: physical dimensions, mass, leg geometry, servo IDs, calibration.
• Hardware_Connections.md: full pinout, power, USART map, RC channels.
• AX12_Control_Table_Mapping.md: AX-12+ control table and buffer offsets.
• Project_History_and_Roadmap.md: project history and phased roadmap (renamed from go-throuugh-all-the-golden-sketch.md).

---

🛠️ Key Firmware Specifications

• Default Balance Gains: $K_p = 95.0$, $K_i = 670.0$, $K_d = 1.9$.
• Loop Rate: Enforced 100 Hz (10,000 µs period).
• Non-Blocking Telemetry RX: Dynamic chunking ($\text{avail}/5$, max 20 bytes per tick) over Serial3 using pipe (|) delimiter.
• Encoder Smoothing: Exponential Moving Average (EMA) velocity filter ($\alpha = 0.15$).
• Servo Half-Duplex Hack: 10 kΩ resistor between PA2 and PA3 with automatic 8-byte echo rejection.