Imagine building your own robot, printing many of its parts on a 3D printer, then teaching it to fold your clothes.
That is the idea behind Sourccey, a new open source home robot created by US startup Vulcan Robotics.
Sourccey can move around a room, see its surroundings and use two robotic arms to interact with objects. It can even learn tasks by watching a human demonstrate what to do.
Perhaps most excitingly for students, makers and schools, the designs are being made available for people to explore themselves.
A robot you can print
Sourccey stands just over one metre tall and weighs around 16 kilograms.
Rather than using an expensive custom made shell, much of its body can be made using a desktop 3D printer and PLA plastic.
This makes the robot easier to experiment with. If a part breaks, it may be possible to print another one. If somebody has an idea for improving a component, they can change the design and print a new version.
It is an approach that has transformed how people develop robots.
In the past, producing a new mechanical part could take days or weeks. With 3D printing, a designer can sometimes create, test and replace a part within hours.
What’s inside Sourccey?
Sourccey might look friendly and fairly simple from the outside, but there is plenty of technology packed inside.
Four special wheels allow it to move forwards, backwards, sideways and diagonally. These are called mecanum wheels.
The robot also has two arms. Each arm has several joints and ends in a gripper that can hold objects.
Four cameras help Sourccey see what is happening around it. A LiDAR sensor can measure the environment using light. Similar technology is used by many other robots and autonomous machines.
At the centre is a Raspberry Pi 5 computer.
This small computer coordinates the robot’s cameras, movement and other systems. Raspberry Pi computers will already be familiar to many schools and computing clubs.
Teaching instead of programming every movement
One of the most interesting things about Sourccey is how it can learn.
Traditional robots are often programmed with detailed instructions telling them exactly how to move.
That becomes difficult with something like clothing.
A T shirt does not always land in exactly the same position. Sleeves bend and fabric wrinkles. Even two identical shirts can look completely different when dropped onto a table.
Sourccey uses a different approach.
A person can control the robot and demonstrate a task. The movements and camera information can then be recorded and used as training data for an AI model.
The robot can use what it has learned to attempt the task itself.
This approach is known as imitation learning.
Learning to tackle the laundry
Laundry provides an interesting challenge for a robot because fabric is unpredictable.
Sourccey has been trained to fold several types of clothing, including T shirts, shorts, jeans and long sleeved shirts.
The project began with some remarkably fast experimentation.
Founder Nick Maselli previously demonstrated a working Sourccey prototype that was assembled in less than 24 hours. Hardware was constructed during the day and AI training took place overnight.
It was not a perfectly polished machine. Parts needed changing and some components had to be printed again.
That is actually an important part of engineering.
Building something, discovering what does not work and improving it is often how new technology develops.
What does open source mean?
Sourccey is particularly interesting because Vulcan Robotics is making the project open source.
Open source means people can look inside a project rather than simply receiving a finished product that they cannot change.
Sourccey’s hardware designs, electronics and software are being made available through GitHub. The STL files used to 3D print parts are also being shared.
That could allow makers, researchers and students to study how the robot works and potentially develop their own ideas.
Someone might create a different gripper. Another person could experiment with a new sensor. Others could work on completely different tasks for the robot to learn.
It turns a robot into a platform for experimentation.
Robotics is becoming more accessible
Sourccey also demonstrates how dramatically robotics has changed.
Many of its technologies would once have been found mainly inside university laboratories and large technology companies.
Today, powerful small computers, affordable cameras, 3D printers and increasingly accessible AI tools are putting those capabilities into the hands of smaller teams and individual makers.
That does not mean building a capable robot is suddenly easy.
Robotics still brings together programming, electronics, mechanics, mathematics, design and problem solving. AI also needs carefully selected training data and testing before it can be trusted to perform tasks reliably.
But the tools needed to start experimenting are becoming much easier to obtain.
Learning from robots that learn from us
There are also important ethical lessons.
A robot learning from human demonstrations depends on the examples it receives. Poor training data can produce poor results.
That becomes increasingly important as robots move from laboratories into homes, schools, hospitals and workplaces.
Developers need to think about safety, privacy and what happens when an AI system makes a mistake.
Open source projects can help people explore these issues because they provide an opportunity to understand what is happening behind the machine.
Sourccey is therefore interesting for more than its ability to fold a T shirt.
It represents a changing world of robotics where people can build, modify, teach and share increasingly capable machines.
The household robot of the future might not arrive as a mysterious machine sealed inside a box.
You might be able to print parts of it, open its code and teach it something completely new.
