DIY Your Own Mini Boston Dynamics Dog with Raspberry Pi

Source: Quantum Bit

What to do if you can’t afford a Boston Dynamics robot dog that costs 530,000 yuan?

How about using an open-source project to DIY one with a Raspberry Pi for less than 4,000 yuan?

With this setup, don’t you feel like you have a Boston Dynamics dog?

DIY Your Own Mini Boston Dynamics Dog with Raspberry Pi

First, let’s warm up and demonstrate the flexible joints:

With three-axis body posture control, this Raspberry Pi version of the Boston Dynamics dog can achieve two modes of motion.

One is the jogging posture. In this quadrupedal gait, the legs diagonally opposite to each other move simultaneously, allowing for faster speeds.

The other is the walking posture. This eight-phase gait, which combines body movement, helps maintain the balance and stability of the robot dog.

This robot dog was created by a foreign guy Michael Romanko.

Are you feeling a bit excited?

Then continue reading to get this detailed open-source tutorial and owning your mini Boston Dynamics dog will not be a dream. (Dog head)

Hardware costs less than 4,000 yuan

This Raspberry Pi robot dog is built on the open-source ThingVerse Spot Micro framework. The shell part is made using a 3D printer with a 0.6 nozzle.

The software part is implemented on a Raspberry Pi 3B computer running Ubuntu 16.04.

So first, you need to prepare a Raspberry Pi 3B development board worth about 230 yuan.

The control components include a servo driver PCA9685 priced at less than 30 yuan, with an I2C interface.

And 12 servo motors PDI-HV5523MG, each priced at about 70+ yuan.

Additionally, you need to equip a 16×2 I2C interface LCD display.

For the battery, a 2s 4000mAh lithium battery is used, directly connected to the power servo, priced around 160 yuan.

The voltage regulator used is HKU5 5V/5A UBEC, which powers the Raspberry Pi, LCD panel, and PCA9695 control board, costing approximately 45 yuan.

The entire setup, including the 3D printer, costs less than 4,000 yuan, even lower than the cost of Stanford’s low-cost open-source quadruped robot Pupper (600-1000 USD).

△ Stanford Pupper

Detailed tutorial down to every detail

Mike’s assembly tutorial is quite brief, but he provides a reference to a community called “Micro Robot Dog AI” (SpotMicroAI).

The SpotMicroAI tutorial is truly “hand-holding”, with the engineer who wrote the tutorial providing detailed steps for assembling the robot dog, even including how to SSH into the Raspberry Pi, ensuring that everyone can understand.

First, install the operating system on the Raspberry Pi, it is recommended to install PINN, which is an enhanced version of the NOOBS operating system installer.

After installation, the system comes with SSH and VNC, so there is no need to connect a keyboard and mouse to the Raspberry Pi, and you can debug it remotely.

△ Successfully connected to the Raspberry Pi via VNC

In the many operating system options of PINN, choose to install Raspbian Lite. After this, disconnect the VNC connection and connect to the Raspberry Pi entirely via SSH.

The Raspberry Pi system comes with FTP. To facilitate beginners in transferring files between the computer and the Raspberry Pi, it is recommended to install FileZilla software.

Enter the Raspberry Pi’s IP address, username, password, port 22, and click Quickconnect to connect to the Raspberry Pi’s file system. Now, you can drag and drop files from SpotMicroAI as needed.

Next, we enter the most crucial step, installing the robot dog control software. To simplify the installation, the author has written a script that automates the installation process:

ssh [email protected] https://gitlab.com/custom_robots/spotmicroai/basic-runtime/-/raw/master/utilities/self_install.sh | bash

Then start the I2C communication bus. Open the terminal and run the following command:

sudo raspi-config

Then make the following changes from the display options: Interfacing options → Enable I2C

Next, connect all the hardware to the Raspberry Pi as shown in the image below:

The 3D printing files for the robot dog’s shell are also provided. The author recommends using a 0.6mm nozzle with a layer height set to 0.3mm to maximize printing speed.

After completing the hardware assembly, it’s time for software debugging.

First, download the default configuration file in JSON format: https : //gitlab.com/custom_robots/spotmicroai/basic-runtime/-//blob/master/spotmicroai.default and place it in the user’s home directory.

Open the calibration tool and calibrate the 12 servos to find the initial position of the motors when the robot dog is at rest.

cd ~/spotmicroai/calibration./calibration.sh

After calibration, the robot dog can perform various actions.

Wait, what is the display screen mounted on the back of the robot dog for? It helps us quickly see the current status of the robot dog.

For example, on the panel above, we can see the current CPU temperature of the Raspberry Pi, the connection status with the external controller, the status of the servo control board, etc.

Thus, the mini robot dog is completed. If you want to assemble it and learn more advanced development tutorials, you can check the content in Extra.

Next Steps

Currently, Mike’s mini robot dog can only perform simple actions. Mike’s next plan is to make it as intelligent as Boston Dynamics’ robot dog. Mike’s future goals are:

1. Add a laser radar to achieve simple 2D mapping of the room through SLAM, just like the original Boston Dynamics robot dog.

Now someone has already implemented Mike’s first goal:

2. Develop an automatic movement module to guide the robot to perform simple tasks around the sensed 2D environment, such as navigating a room and avoiding obstacles.

3. Create software modules with cameras or webcams for basic image classification. For example, perceiving a fist clenching and unclenching, allowing the robot to respond to specific actions for gesture control.

Netizens: Let’s DIY

DIY your own “Boston Dynamics dog”, such practices have even received praise from Raspberry Pi’s official account.

Netizens have expressed: “This is really cool.”

Some are eager to start making it themselves.

So, shall we DIY?

Reference Links:

https://www.raspberrypi.org/blog/mini-raspberry-pi-boston-dynamics-inspired-robot/

https://github.com/mike4192/spotMicro#Overview

https://gitlab.com/custom_robots/spotmicroai

- end -


Lastly, here is a great Pandas documentation. Pandas is a powerful toolkit for analyzing structured data; it is based on Numpy (which provides high-performance matrix operations); it is used for data mining and data analysis, and also provides data cleaning functions. The best way to learn Pandas is to read the official documentation:

"10 Minutes to pandas", "Pandas cookbook", "Learn Pandas". The English version can be a bit challenging, so I recommend the Chinese version of the official Pandas documentation!



Scan the QR code below and reply with 'pandas' to get the download link.


Silly Programmer