AB
AiBoss
project

RoboOS 2.0 - Zhipu Open Source's cross-ontology cerebellum-cerebellum collaborative framework

RoboOS 2.0 is an open-source, cross-ontology brain-brain collaborative framework developed by Zhipu, specifically designed for embodied intelligence. The framework supports multi-robot collaboration, achieving lightweight deployment based on the integrated MCP protocol and serverless architecture, thus lowering the development threshold.

What is RoboOS 2.0?

RoboOS 2.0 is an open-source, cross-entity cerebellum-cerebellum collaborative framework designed specifically for embodied intelligence. The framework supports multi-robot collaboration, achieving lightweight deployment based on the integrated MCP protocol and serverless architecture, lowering the development threshold. The framework includes a cloud-based brain module responsible for advanced cognition and multi-agent collaboration; a distributed cerebellum module group dedicated to executing specific robot skills; and a real-time shared memory mechanism to enhance environmental situational awareness. RoboOS 2.0 provides standardized interfaces to eliminate hardware compatibility differences and uses a skill store to achieve intelligent matching and one-click adaptation of robot skill modules, helping robots move from "single-machine intelligence" to "swarm intelligence."

Main features of RoboOS 2.0

  • Multi-robot collaborationIt supports dynamic allocation and parallel execution of multi-agent tasks, is suitable for complex scenarios, and improves task execution efficiency.
  • Cerebellum-Cerebellum CoordinationThe brain module is responsible for advanced cognition and multi-agent collaboration, while the cerebellum module is dedicated to the execution of robot-specific skills, achieving efficient division of labor.
  • Lightweight deploymentIt integrates the MCP protocol and serverless architecture, lowering the development threshold, supporting rapid deployment, and simplifying the development process.
  • Standardized InterfaceIt provides standardized interfaces to eliminate compatibility differences between different manufacturers and hardware, and supports one-click adaptation of robot skill modules created by developers worldwide.
  • Real-time perception and modelingA new multi-ontology spatiotemporal memory scene graph sharing mechanism has been added, which supports real-time perception and modeling in dynamic environments and enhances environmental adaptability.
  • Task monitoring and feedbackA multi-granularity task monitoring module is introduced to achieve closed-loop task feedback, improve the stability and success rate of task execution, and ensure reliable task completion.

Technical Principles of RoboOS 2.0

  • Hierarchical task decompositionComplex tasks are broken down into sub-tasks and dynamically allocated through network topology to ensure efficient task execution.
  • End-to-cloud collaboration:
    • Brain cloud-based optimized reasoning deploymentLeveraging the powerful computing capabilities of cloud computing, it enables advanced cognition and multi-agent collaboration.
    • Cerebellar skills' no-adaptation registration mechanismIt supports rapid deployment of cerebellum modules and skill registration, significantly reducing the development threshold.
  • Real-time shared memory mechanismBased on a real-time shared memory mechanism, the environment status and task progress are dynamically updated to ensure efficient collaboration among multiple agents.
  • Multimodal data processingIt supports the processing of multimodal data such as high-resolution images, multi-view videos, and scene graphs, thereby improving the model's perception and reasoning capabilities.
  • System-level optimizationThe system-level optimization of the end-to-end inference link improves overall performance by up to 30%, increases edge-cloud communication efficiency by 27 times, and reduces the average response latency of the entire link to less than 3ms.

The project address for RoboOS 2.0

  • Project official websitehttps://github.com/FlagOpen/RoboOS
  • GitHub repositoryhttps://github.com/FlagOpen/RoboOS
  • arXiv technical paper: https://arxiv.org/pdf/2505.03673

Application scenarios of RoboOS 2.0

  • Supermarket LogisticsMultiple robots collaborate to complete tasks such as goods handling and shelf organization, with dynamic path planning and real-time obstacle avoidance, improving logistics efficiency.
  • Home servicesRobots assist with household chores, such as cleaning and tidying up, and can sense changes in the environment in real time to adapt to dynamic home scenarios.
  • Industrial productionMultiple robots work collaboratively on the production line to complete the tasks of handling and assembling parts, thereby improving production efficiency and quality.
  • Medical careRobots assist in nursing work in hospitals, such as carrying medical supplies and helping patients move around, reducing the burden on medical staff.
  • Public facilities maintenanceRobots collaborate to complete tasks such as cleaning public areas and inspecting equipment, providing real-time status feedback to ensure the normal operation of facilities.