Demonstrating Berkeley Humanoid Lite: An Open-source, Accessible, and Customizable 3D-printed Humanoid Robot

Yufeng Chi, Qiayuan Liao, Junfeng Long, Xiaoyu Huang, Sophia Shao, Borivoje Nikolic, Zhongyu Li, Koushil Sreenath

Seminal April 2025

The robot walks under the RL locomotion policy, following velocity commands.

TL;DR An open-source, 0.8 m, 16 kg humanoid built from desktop-3D-printed parts and off-the-shelf components for under $5,000. Its 3D-printed cycloidal actuators survive a 60-hour durability test, and the robot walks with an RL policy transferred zero-shot from simulation.

What they do

Most humanoids are either expensive closed commercial products or lab prototypes that need CNC machining. Berkeley Humanoid Lite is a mid-scale humanoid (0.8 m, 16 kg) that anyone can build: every non-standard part is printed in PLA on a desktop FDM printer, everything else comes from common online vendors, and the hardware, embedded code and training and deployment frameworks are open-source.

How it works

3D-printed cycloidal actuators. Two actuator sizes (6512 and 5010) pair a drone BLDC motor and a low-cost motor driver with a printed cycloidal gearbox. Cycloidal gears spread load over many teeth and tolerate the limited resolution of desktop printers better than planetary gears; through-screws and embedded brass hex standoffs reinforce the layer boundaries.
Simple system. Each joint is a self-contained actuator needing only power and CAN. An Intel N95 mini PC talks to four CAN buses and an IMU at 250 Hz; a 6S 4000 mAh battery gives about 30 minutes of operation.
Customizable. With no linkages, link lengths and hip joint order are easy to change, and the same actuators are rearranged into a quadruped, a biped, a centaur and a mobile base.
Control. A PPO policy trained in Isaac Gym on proprioception runs onboard at 25 Hz; the arms are teleoperated with SteamVR controllers and an inverse-kinematics solver.

Results

Bill of materials: $4,312 in the US and $3,236 in China; a 6512 actuator costs $188. The robot can be assembled in about three days.
Gearbox mechanical efficiency is about 90% across most operating conditions. Over a 60-hour durability test, efficiency dipped and recovered while backlash rose slightly.
Six actuators from two printers track torque within ±0.5 Nm; maximum backlash is 0.0229 rad. A 5-DoF arm repeats target positions with a 3.433 mm standard deviation.
The walking policy transfers zero-shot and uses only 30% of the actuator torque limit. Teleoperation covers writing, packing a box, picking blocks and solving a Rubik’s Cube.

Limitations

Thermal effects on the printed structure during prolonged operation are not studied, and efficiency already drops at high torque and speed from heat. Transmission stiffness (about 319 Nm/rad) is well below the 1468 Nm/rad reported for a similar reducer printed in carbon-fiber polyamide.

Videos

Teleoperated bimanual manipulation of a Rubik's Cube.

Teleoperated writing with a marker.

Figures

From Figure 2: Main components and key dimensions of the limbs of the Berkeley Humanoid Lite.

From Figure 4: Exploded and cross-sectional views of the 6512 actuator. All of the components are either 3D-printed o... — From Figure 4: Exploded and cross-sectional views of the 6512 actuator. All of the components are either 3D-printed or sourced from online marketplaces.

From Figure 3: Comparison between existing robot platforms. Y-axis shows the performance factor, the average peak tor... — From Figure 3: Comparison between existing robot platforms. Y-axis shows the performance factor, the average peak torque normalized by the weight and height of the robot. A higher performance-per-dollar (dotted contours) indicates better cost-effectiveness.

From Figure 15: Snapshot of teleoperated manipulation experiments, including four tasks: writing, unpacking and packi... — From Figure 15: Snapshot of teleoperated manipulation experiments, including four tasks: writing, unpacking and packing, picking and placing blocks, and solving Rubik’s Cube.

Abstract

Despite significant interest and advancements in humanoid robotics, most existing commercially available hardware remains high-cost, closed-source, and non-transparent within the robotics community. This lack of accessibility and customization hinders the growth of the field and the broader development of humanoid technologies. To address these challenges and promote democratization in humanoid robotics, we demonstrate Berkeley Humanoid Lite, an open-source humanoid robot designed to be accessible, customizable, and beneficial for the entire community. The core of this design is a modular 3D-printed gearbox for the actuators and robot body. All components can be sourced from widely available e-commerce platforms and fabricated using standard desktop 3D printers, keeping the total hardware cost under $5,000 (based on U.S. market prices). The design emphasizes modularity and ease of fabrication. To address the inherent limitations of 3D-printed gearboxes, such as reduced strength and durability compared to metal alternatives, we adopted a cycloidal gear design, which provides an optimal form factor in this context. Extensive testing was conducted on the 3D-printed actuators to validate their durability and alleviate concerns about the reliability of plastic components. To demonstrate the capabilities of Berkeley Humanoid Lite, we conducted a series of experiments, including the development of a locomotion controller using reinforcement learning. These experiments successfully showcased zero-shot policy transfer from simulation to hardware, highlighting the platform's suitability for research validation. By fully open-sourcing the hardware design, embedded code, and training and deployment frameworks, we aim for Berkeley Humanoid Lite to serve as a pivotal step toward democratizing the development of humanoid robotics. All resources are available at https://lite.berkeley-humanoid.org.

Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.

Matt Strong