Shanghai, October 15th / PRNewswire / — Today, RP1 ( ROBOTO 01) is officially released. This is a high-performance, full-stack open-source humanoid robot developed for advanced robotics research and embodied AI. The hardware and software are openly accessible, allowing researchers to customize the system, integrate their own algorithms, and test new methods on real hardware.
As AI increasingly enters the physical world, humanoid robots are becoming an important platform for the next generation of embodied intelligence. But if their core technologies remain encapsulated in black boxes, will innovation be defined only by a few who have access to the underlying systems?
The future of embodied AI should not be built upon technological black boxes. Every researcher and developer should have the opportunity to be a part of it.
The AI model continues to improve, but there are still significant limitations when it comes to applying its capabilities to real-world physical systems.
For robotics researchers, closed hardware and underlying control systems limit the flexibility of experiments and system innovation. For developers of embodied AI, the limited visibility of the internal behavior of robots makes it difficult to pinpoint issues between model strategies, control systems, and hardware execution.
RP1 addresses these limitations by making its robot ontology design and technical resources available to the public. These include the ROMOMO joint actuator module developed in-house by RoboParty, as well as technical documentation and research materials. Together, these resources have contributed to the establishment of an open development ecosystem that encompasses mechanical structures, actuators, motion control, simulation environments, software development toolkits ( SDK ), and training tools.
Full open-source for both hardware and software
Open-source robots should not be limited to the software interface level alone. To advance research in humanoid robots, developers need to understand, modify, and verify the physical systems that support intelligent behavior.
RP1 is precisely built around this principle.
At the hardware level, RP1 integrates the ROMOMO joint actuator module developed by RoboParty, as well as an optimized mechanical structure and motion control architecture. Its underlying drive and control algorithms are also completely open-source. The software stack extends to firmware, development tools, simulation resources, and robot learning frameworks. By making the hardware design and underlying technologies open, researchers are able to explore the relationships between mechanical design, joint response, control behavior, and learning strategies.
Researchers no longer just send instructions to a fixed robot platform; they can also analyze and modify the underlying systems that determine how the robots perform their actions.
When a robot fails to perform the expected action accurately, the issues may lie in the model, control strategy, actuator response, or mechanical structure. An open development ecosystem enables researchers to track problems across these aspects and collaborate to optimize on a verifiable and modifiable system basis.
True open source should not only extend to the code, but also to the robot itself, handing over full control of the system to the developers.
Designed for research needs
The standard version RP1 has a height of about 1.3 meters and possesses 24 drivable degrees of freedom (DoF), supporting gait, balance control, and simulation-to-reality (sim-to-real) migration research. The Pro version features 30 drivable degrees of freedom, including 2 degrees of freedom for the neck and two robotic arms with 7 degrees of freedom each, supporting end-to-end strategies, human-computer interaction, and embodied AI data collection research. The aforementioned number of degrees of freedom does not include the dexterous hand.
The modular structure and open control interfaces of RP1 enable the research team to adjust the platform according to experimental needs. The team can start with research on bipedal walking and then configure additional hardware and software for perception, data collection, or experimental operations.
Interchangeable head modules and wrist end-effectors support a variety of configurations, including the RP Hand dexterous hand developed by RoboParty. These interfaces enable teams to conduct a range of experiments on the same robotic platform.
Open hardware and software also support controlled experiments. Researchers can compare different control methods on the same hardware, or observe how changes in hardware configuration affect experimental results while keeping the control methods unchanged. Recording hardware versions, software configurations, and test conditions helps teams trace the causes of differences in results.
Help developers focus on embodied AI models
RP1 now supports basic motor abilities, including continuous walking, recovery from external disturbances, getting up after falling, and tracking human movements. These capabilities provide a ready-made foundation for the development of embodied AI models, reducing the need for teams to repeatedly develop underlying motor skills. This allows them to focus on higher-level intelligent strategies and complex tasks.
In terms of whole-body intelligent development, RP1 provides a robot body that is capable of dynamic movement and whole-body control (WBC), as well as tools that support virtual reality (VR) teleoperation, data collection and playback, and action redirection. It also offers an integrated development process for whole-body action execution based on WBC, teleoperation through VR or motion capture devices, and control via RP Hand.
WBC coordinates the arms, torso, and legs to support dynamic motion tracking and posture adjustment, providing a stable foundation for action execution in tasks that combine walking and manipulation. Combined with RoboParty's self-developed PartyOS embodied AI system, UFO unsupervised reinforcement learning framework, and MimicLite action imitation technology, RP1 supports the learning of motor skills, the development of control strategies, and higher-level applications.
RP1 also provides an open toolchain that covers data collection, simulation training, physical robot deployment, and experimental verification. Researchers can deploy models onto robots, analyze the differences between model outputs and actual physical responses, and independently adjust control strategies and hardware configurations. This forms a continuous development loop from algorithm development to real-world execution, feedback, and optimization.
To meet the requirements for experimental stability and repeatability in the development of the embodied AI, RP1 has made systematic improvements in mechanical structure, core actuators, and manufacturing processes. The ROMOMO actuators developed in-house by RoboParty include structural optimizations and manufacturing quality management, aimed at enhancing physical response consistency, overall reliability, and maintainability. These measures help to minimize the impact of hardware differences on experimental results and provide a stable physical platform for long-term repetitive training and validation.
Build a shared foundation for global innovation
RoboParty launched the RPO open-source project in January 2026, releasing mechanical and electrical design resources, as well as training and deployment materials. The project has received over 2600 GitHub stars, enabling developers to reproduce and expand robot designs using these public resources.
RP1 continued this open-source approach by integrating the ROMOMO joint actuator module, the research findings of RoboParty Lab, and the ongoing development of PartyOS to expand the technical foundation for robot research and the development of embodied AI models.
RoboParty founder Yi Huang stated: "We believe that the future of embodied AI should not be defined by a few closed systems. Through RP1, we aim to open up more than just the code. We want researchers to truly understand, modify, and continuously advance robot capabilities. When more teams can innovate on a shared technical foundation, the entire industry can go even further."
RoboParty represents a vision to make the future of embodied AI accessible to everyone who is ready to explore it.
GitHub : https :// github.com / Roboparty / RP1
Website: https :// roboparty.com
Media contact: [ email protected ]
About RoboParty
RoboParty was established in 2025, with its headquarters located in Shanghai. The company develops humanoid robot hardware, motion control technology, robot software, and open development tools. Its product and technology portfolio includes the RPO and RP1 humanoid platforms, self-developed ROMOMO joint actuator modules, the RP Hand dexterous hand, as well as the continuously evolving PartyOS embodied AI system. By providing publicly available engineering resources and development tools, the company supports robot research, embodied AI development, and collaboration among teams.
