- Published on
China’s RoboParty Launches Open-Source RP1 Humanoid for Robotics Research
- Authors
- Name
- DeepHorus
Shanghai-based RoboParty launched RP1, a full-stack open-source humanoid robot for robotics research and embodied AI. It gives developers access to hardware and software, not a closed interface. RP1 is available in Standard and Pro versions and through GitHub.
RP1 extends RoboParty’s RPO open-source project, launched in January. RPO released mechanical and electrical designs, training, and deployment materials. RoboParty said the project has received more than 2,600 stars on GitHub.
Standard RP1 is about 1.3 meters tall with 24 actuated degrees of freedom, targeting gait, balance, and sim-to-real research. RP1 Pro has 30 actuated degrees of freedom, including a two-degree-of-freedom neck and two seven-degree-of-freedom arms. Both support modular hardware, interchangeable head modules, and the RP Hand dexterous hand.
The robot supports continuous walking, recovery from disturbances, standing after falls, and human-motion tracking. Researchers can use virtual-reality teleoperation, motion capture, data collection and playback, motion retargeting, and whole-body control. Whole-body control coordinates arms, torso, and legs for combined locomotion and manipulation.
The platform includes ROMOMO joint actuators, low-level drivers, control algorithms, firmware, simulation tools, SDKs, and robot-learning frameworks. This lets researchers trace failures across the full system, from AI model and control policy to actuator response and mechanical design. It also works with PartyOS, the UFO reinforcement-learning framework, and MimicLite motion-imitation technology.
“The future of embodied AI should not be defined by a handful of closed systems,” founder Yi Huang said. He said RP1 aims to let researchers understand, modify, and advance robotic capabilities. RoboParty also stressed consistency across mechanical structure, actuators, and manufacturing so experiments can be repeated.
The development workflow spans data collection, simulation training, deployment on physical robots, and testing. Researchers can compare model outputs with actual behavior and adjust control policies or hardware. By opening hardware and software, RoboParty hopes more teams can innovate and advance the industry.
