Mini-Quadruped
B✓ Links überprüft Jul 16, 2026
Nur Anleitungen und Stückliste – die orobot-Firmware unterstützt diese Hardware noch nicht.
Zum Erkunden anmeldenÜber dieses Programm
Mini-Quadruped
A low-cost open-source quadruped robot platform built around an NVIDIA Jetson Nano (vision + ML) and an STM32F103C8T6 'Blue Pill' microcontroller (gait + inverse kinematics). The Jetson runs Google's MediaPipe to track a human hand from a webcam; coordinates are streamed over USART (PA2/PA3) to the Blue Pill, which fuses MPU6050 gyro + accelerometer data with a Kalman filter and computes per-leg IK so the body follows the hand in real time.
Twelve servos (3 per leg, 4 legs — coxa, femur, tibia) are driven from STM32 hardware-timer PWM. Body orientation is closed-loop on the IMU.
What you'll learn
- Jetson Nano + MediaPipe vision pipeline
- STM32 servo control via hardware-timer PWM
- Inverse kinematics for a 3-DOF quadruped leg
- Sensor fusion (Kalman) on MPU6050 IMU
- USART link between a Linux SBC and a bare-metal MCU
Hardware (BYOD)
This is a bring-your-own-hardware build. It does NOT run on the standard orobot Raspberry Pi firmware. You'll need:
- NVIDIA Jetson Nano (vision host, MediaPipe)
- STM32F103C8T6 Blue Pill (servo + IK + IMU fusion)
- 12x Afrc D2307HC 7.5kg servos
- MPU6050 IMU, 2x XL4005 buck converters, USB webcam, 31 3D-printed parts
Source
- GitHub: https://github.com/nicoRomeroCuruchet/mini-quadruped
- Author: Nico Romero Curuchet (@nicoRomeroCuruchet)
- Demo videos: https://youtu.be/wdcIH24sI44 https://youtu.be/hFUbpantrfs
The control code in this program is a learning stub — the buttons (Stand / Sit / Walk / Tilt / Follow Hand) wire up to placeholder handlers showing how a UI talks to the real STM32 firmware over USART. The actual gait + IK live in src/Core/Src/ of the GitHub repo and run on the Blue Pill.
🖨 Druckdateien (15)
base_chasis_v1.stl
front_base_v1.stl
rear_base_v1.stl
side_chasis_v1.stl
left_front_coxa_v1.stl
left_front_femur_v2.stl
Benötigte Hardware
| Item | Qty | Notes |
|---|---|---|
| NVIDIA Jetson Nano | 1 | Vision host — runs MediaPipe hand tracking (Python) |
| STM32F103C8T6 Blue Pill | 1 | Servo + IK + MPU6050 fusion (bare-metal C, STM32CubeIDE) |
| Afrc D2307HC 7.5kg servo | 12 | 3 per leg × 4 legs (coxa, femur, tibia) |
| MPU6050 IMU | 1 | Gyro + accelerometer, I2C1 on PB6/PB7 |
| XL4005 5A buck converter | 2 | DC-DC step-down for servo + logic rails |
| Experimental perfboard 20×80 mm | 2 | Wiring harness |
| USB webcam | 1 | Hand tracking input on Jetson Nano |
| 3D-printed structural parts | 31 | Chassis (7) + 4 legs × 6 parts each — see STL files |
Ähnliche Roboter entdecken