RynnRCP - An open-source robot context protocol from Alibaba DAMO Academy

What is RynnRCP?

RynnRCP is an open-source Robotics Context Protocol from Alibaba DAMO Academy that streamlines the entire embodied intelligence development process. RynnRCP comprises two main modules: the RCP framework and RobotMotion. The former provides standardized interfaces for the robot body and sensors; the latter acts as a bridge between cloud inference and robot control, converting low-frequency inference commands into high-frequency control signals. Through standardized protocols and tools, RynnRCP lowers the development threshold and facilitates the efficient adaptation and implementation of embodied intelligence from data acquisition to action execution.

Main functions of RynnRCP

• RCP framework:
  • Capability AbstractionIt provides abstractions of the robot body and sensor capabilities, encapsulating complex hardware interfaces into standardized service interfaces for easy use by developers.
  • Multi-protocol supportIt supports multiple communication protocols (such as MQTT, WebSocket, LCM, etc.) to enable efficient communication between robots and cloud platforms and edge devices.
  • Modular designDevelopers can extend and customize service nodes according to their needs, such as implementing functions like device occupancy control, resource scheduling, and multi-client collaboration.
  • secure communicationThe configuration file only stores device authentication metadata. At runtime, it generates time-sensitive access tokens through an HTTPS secure channel to ensure secure communication.
• RobotMotion:
  • Low-frequency to high-frequency conversionIt converts discrete low-frequency inference commands into high-frequency continuous control signals in real time, ensuring the smoothness and consistency of robot motion.
  • Simulation and debugging toolsIt provides a physics simulation tool based on MuJoCo, which supports motion planning and verification in the simulation environment and reduces the difficulty of policy transfer.
  • Data Acquisition and PlaybackIt supports data acquisition and playback functions, making it convenient for developers to perform visual analysis of robot motion trajectories.
  • Real device debuggingIt provides real-machine debugging capabilities, enabling rapid verification and optimization of control strategies on actual robots.
• Camera Node:
  • Real-time image acquisitionThis project uses OpenCV to implement real-time image acquisition from multiple cameras, supporting dynamic adjustment of resolution and frame rate.
  • Lossless compressionBased on Gzip, lossless compression of the original image is performed, reducing network transmission bandwidth consumption.
  • Asynchronous processingBy leveraging Python's multithreading capabilities, image acquisition and message response are separated, ensuring system responsiveness and resource utilization efficiency.

RynnRCP Technical Principles

• Robot Context Protocol (RCP)RCP is a standardized communication protocol used to define the interaction between the robot body, sensors, and cloud platform. It encapsulates hardware interfaces into generic service interfaces through an abstraction layer, enabling seamless integration between different hardware and models. RCP supports multiple communication protocols (such as MQTT, WebSocket, LCM, etc.) and uses an adaptation layer to achieve protocol conversion and compatibility, ensuring efficient and stable data transmission. It defines the content and structure of data transmission using standardized data formats (such as Protobuf and LCM message types), making it easy for developers to understand and use.
• Modular designThe RCP framework uses a modular design, dividing robot services into multiple independent modules (such as ActionServer, SensorServer, DeviceMonitorServer, etc.), each responsible for specific functions. Developers can extend and customize these modules according to their needs. It provides a unified development paradigm and basic modules, making it easy for developers to get started quickly and reducing development costs.
• Low-frequency to high-frequency conversionThe RobotMotion module uses real-time control algorithms to convert discrete low-frequency inference commands into high-frequency continuous control signals, ensuring the smoothness and consistency of robot motion. Combined with physical simulation tools (such as MuJoCo), it plans and optimizes robot motion to ensure the trajectory conforms to physical constraints.

RynnRCP's project address

• GitHub repositoryhttps://github.com/alibaba-damo-academy/RynnRCP

Application scenarios of RynnRCP

• Industrial AutomationThrough standardized protocols and real-time control technology, precise control and task execution of robotic arms on industrial production lines can be achieved, thereby improving production efficiency and product quality.
• Logistics warehousingIn a logistics warehouse, AGVs and robots are controlled to complete the tasks of handling and sorting goods, while the inventory status is monitored in real time to optimize the logistics process.
• service robotsIt supports service robots in homes, hotels, restaurants, and other settings to perform tasks such as cleaning, food delivery, and providing information, thereby improving service efficiency and user experience.
• Medical RehabilitationIt is used to control rehabilitation robots, adjust training intensity according to the patient's rehabilitation progress, and support surgical assist robots to provide high-precision surgical support.
• Agricultural and Environmental MonitoringIt controls agricultural robots to complete tasks such as sowing, irrigation, and harvesting, while also being used in environmental monitoring robots to collect and upload monitoring data in real time, thus contributing to agricultural production and environmental protection.