The company is a fast-rising pioneer on China's humanoid robot scene. With deep links to the country's prestigious Tsinghua University, it aims to lead the way in the research and development of general-purpose robotics with the potential to solve real-world problems and benefit mankind.

Beijing, China, March 12 -- Robot Era, a fast-rising humanoid robot startup, has announced the decision to open-source its reinforcement learning (RL) framework Humanoid-Gym on GitHub.

The move was intended to lower the threshold to developing RL algorithms for humanoid roboticists worldwide, providing them with a practical platform to build on to avoid "reinventing the wheels."

Robot Era, a Beijing-based company, is one of the first players in the industry worldwide to showcase a humanoid robot's locomotion ability based on end-to-end reinforcement learning.

In videos posted earlier, the company's full-size humanoid robot, XBot-L, standing 1.65 meters tall, can be seen walking on slippery snow and climbing up and down stairs with stability and dexterity, as part of a test on the robot's locomotion capabilities.

Due to the complex structure and control mechanisms, humanoid robots pose a bigger challenge than quadrupedal robotics in the transition from a simulation avatar to a working prototype in the real world.

This process is known as Sim-to-real policy transfer in the parlance of reinforcement learning.

Based on the specific reward functions and domain randomization signals, the Humanoid-Gym framework from Robot Era can appreciably simplify the steps humanoid robots need to take to achieve Sim-to-real translation.

This could significantly streamline the development process, reducing technical difficulty and bolstering engineering productivity.

The framework was previously put to the test on Robot Era's two models, XBot-S and XBot-L, and proved a success.

Another benefit of Humanoid-Gym is that it is calibrated to allow for proof-of-concept tests on humanoid robots in MuJoCo, an advanced simulator for multi-body dynamics, so as to ensure close alignment between a simulated and a real robot model.

The framework also introduced several variables for the assessment of reinforcement learning results, including but not limited to speed tracking and the smoothness of robots.

The open-sourcing of Humanoid-Gym embodies a strong desire on the part of Robot Era to foster knowledge-sharing and innovative exchange in the area of humanoid robotics.

About Robot Era

Founded in August 2023 and incubated by Tsinghua University's Institute for Interdisciplinary Information Sciences, Robot Era is committed to the development of embodied AI and humanoid robotic technologies and products.

Project page:
https://sites.google.com/view/humanoid-gym/
GitHub page:
https://github.com/roboterax/humanoid-gym