Berkeley Humanoid is a low-cost, easy-to-simulate robot

Researchers at UC Berkeley have released a mid-sized humanoid robot that is safe, affordable, and has a thin sim-to-real gap.

Researchers at UC Berkeley have developed Berkeley Humanoid, a robot platform designed for affordability, modularity, ease of repair, and a tight sim-to-real gap. It is meant to address the high costs and complexity of robotics research.

Berkeley Humanoid is a 16-kilogram, fully electric, mid-scale humanoid robot built almost entirely from custom-designed parts, including actuators, transmissions, sensors, and the control board. Without arms, the Berkeley Humanoid costs around $10,000 to manufacture, which is significantly cheaper than other comparable robots.

It is also safe to operate and can be easily maneuvered and reset by a single operator for indoor environments. 

The simple mechanical design enables the researchers to run simulations of the robot at very high speed. On an Nvidia A4500 GPU, the robot can be simulated at a speed of more than 90,000 steps per second.

The Berkeley Humanoid uses a simple model-free reinforcement learning (RL) model that takes the robot’s sensor data as input and outputs motor commands. According to the researchers, the simple hardware and software design of the Berkeley Humanoid helped reduce the sim-to-real gap.

During tests, Berkeley Humanoid was able to successfully walk on different surfaces, including grass fields, sidewalks, unpaved trails, asphalt roads, and bridges. It was also able to navigate inclines and stairs.

It also has robust recovery capabilities and can regain its balance after being kicked in different directions.

The researchers hope to turn it into a reliable, low-cost research platform that can be deployed scalably for learning in the real world.

Read more about Berkeley Humanoid on TechTalks, including comments from the lead author.

Read the paper on arXiv.