Why Not Build This Bot?

It is an undeniable fact: humanoid robots are cool. It is just one of those things, like how everyone knows that puppies make people smile or that nobody can resist freshly baked chocolate chip cookies. These robots have a built-in awesome factor, in much the same way that sunglasses automatically make anyone that wears them look 25 percent cooler.

For the casual hobbyist, however, getting a reasonably capable humanoid robot in your hands is easier said than done. Commercial systems tend to be both very expensive and complicated, leaving them far out of reach. As a hardware hacker, there is always the possibility of building your own, but that is a task that stretches the abilities of even very seasoned armchair engineers. Building a decent humanoid robot requires advanced knowledge of mechanics, electronics, software engineering, and much more.

That is a big hurdle, to say the least, but all is not lost. There is still another option — borrowing from those that have already done the work and made it open source. Among these projects, you would be hard-pressed to find one better than Zeroth Bot, which is a creation of Jingxiang Mo, Kelsey Pool, and Denys Bezmenov. Their open source robot is capable, hackable, and the parts only cost $350, which is very reasonable in the world of robotics. Best of all, they have released an assembly guide, as well as all of the software and instructions you need to get your own Zeroth Bot up and running.

The body of the robot is 3D-printed and is equipped with a powerful Milk-V DUO S development board sporting a dual-core RISC-V CPU clocked at 1 GHz. An RP2040 MCU Board adds a 1.28-inch round LCD display and an inertial measurement unit. A Milk-V CAM-GC2083 camera installed in the robot’s head enables computer vision applications, and a microphone and speaker provide for more interaction possibilities. A set of 16 servo motors control the robot’s movements, and driver boards and rechargeable batteries round out the hardware components.

Zeroth Bot may be small, but don’t let its size fool you — it is capable of running the latest artificial intelligence control algorithms. It is also a perfect testbed for experimenting with reinforcement learning techniques and even learning from simulated environments with tools like NVIDIA’s Isaac Gym. By experimenting with speech, vision, locomotion, teleoperation, and more on Zeroth Bot, you will have the opportunity to learn many valuable skills that can be transferred to more powerful, commercial robotics systems.

Work on the robot is still actively underway, and the development team is encouraging members of the community to contribute. So if you do make your own Zeroth Bot and do some interesting things with it, be sure to share so that the entire community can benefit. And be sure to post your project to Hackster as well — we would love to see your work!