Project Title: Cube_Mini (2023 - Present)

Project Overview

Cube_Mini is a compact self-balancing cube powered by two ESP32 microcontrollers and three brushless Permanent Magnet Synchronous Motors (PMSMs). Inspired by the CubeMini project on GitHub , this system combines advanced sensor feedback, precise PID control, and real-time wireless communication to dynamically maintain balance in three dimensions. The frame is built with lightweight carbon fiber and reinforced with custom 3D-printed corners for structural integrity.

Objectives

• Self-Balancing Control: Implement real-time PID feedback loops using inertial measurements to actively stabilize the cube in three axes.
• Modular Motor Control: Use distributed microcontroller architecture where each ESP32 manages motor control independently and communicates over WiFi.
• Wireless Parameter Tuning: Enable remote control and dynamic tuning of PID parameters through a custom WiFi TCP protocol for real-time system adjustment and calibration.
• Lightweight and Durable Design: Utilize carbon fiber and 3D-printed components to minimize weight while maximizing mechanical stability.

Hardware Components

• Microcontrollers:
  • 1 x ESP32 (controls 1 PMSM, acts as WiFi host)
  • 1 x ESP32 (controls 2 PMSMs, WiFi client)
• Motors: 3 x Brushless Permanent Magnet Synchronous Motors (PMSMs)
• Sensors: 1 x MPU6050 6-DOF IMU (accelerometer + gyroscope)
• Structural Frame:
  • Carbon fiber rods for the main structural frame
  • Custom 3D-printed corners for motor mounting and frame joints
• Power Supply: Li-Po battery pack and motor driver circuits designed for low-latency operation

Software Configuration

• Motor Control:
  • PID loops implemented for real-time control of motor torque based on tilt angle and angular velocity feedback
  • Sensor fusion from the MPU6050 for stable gravity vector estimation
• Communication:
  • WiFi-based TCP socket communication between ESP32 nodes
  • Host ESP32 acts as control hub, accepting parameter updates via external interface
• Sensor Integration:
  • Real-time data acquisition from the MPU6050
  • Sensor calibration and filtering for noise reduction and accurate orientation data
• Parameter Tuning Interface:
  • Custom TCP protocol allows live PID gain tuning, data monitoring, and diagnostics over wireless connection

Implementation Details

1. Mechanical Design:
   • Designed a minimalistic cube using carbon fiber tubes for weight efficiency and structural strength
   • Modeled and printed 3D-printed motor mounts and corners to support each motor at the cube’s edges
2. Motor and Sensor Wiring:
   • Routed power and control signals through hollow carbon fiber tubes
   • Integrated MPU6050 securely at the geometric center of the cube for optimal inertia measurement
3. ESP32 Control Architecture:
   • Programmed one ESP32 to handle a single-axis control and host TCP server for PID parameter input
   • Second ESP32 handles two axes, receiving instructions from the host node and executing synchronized motor commands
4. Control System Development:
   • Tuned PID loops for each motor based on real-time sensor feedback
   • Developed filtering algorithms to mitigate gyro drift and sensor noise
5. Wireless Protocol Design:
   • Implemented a low-latency TCP interface using ESP-IDF framework for robust two-way communication
   • Created Python-based ground station interface to send PID parameters and monitor system response

Challenges and Resolutions

• Sensor Drift and Noise: Applied Kalman and complementary filters to stabilize accelerometer and gyroscope outputs.
• Motor Sync Timing: Resolved timing jitter between ESP32s by implementing soft time-synchronization protocol over TCP.
• Wireless Latency: Optimized TCP packet size and reduced unnecessary message overhead to ensure fast PID loop updates.
• Frame Resonance: Addressed mechanical resonance by reinforcing joints and tuning PID values to reduce oscillatory behavior.

Outcomes

• Functional Self-Balancing Cube: Achieved dynamic equilibrium on all three axes using a distributed control system and feedback loops.
• Live PID Tuning: Enabled real-time parameter tuning via WiFi, drastically reducing iteration time during testing.
• Modular Hardware Design: Validated the modular architecture, allowing future replacement or scaling of motor/ESP components.
• Educational Value: Gained experience in closed-loop control systems, sensor integration, embedded communication, and system dynamics.

Future Plans

• Closed-Loop System Modeling: Integrate real-time telemetry and Bode plot visualization to enhance control system analysis.
• Mobile App Interface: Develop a user-friendly app for Bluetooth or WiFi tuning without a PC interface.
• Advanced Motion Behaviors: Implement trajectory planning and dynamic rebalancing for hopping, reorientation, and controlled falling.
• Sensor Redundancy: Add magnetometer or secondary IMU for improved spatial awareness and fail-safe control.

This project showcases a high-performance, scalable platform for exploring embedded control systems and real-time dynamics in robotics.