Cyberwave is composed of modular subsystems that work together to manage the robotic lifecycle.
Core Modules
- Digital Twin Engine: The heart of the platform. It ingests URDF files to create kinematic models and synchronizes them with real-time telemetry. This allows for in-browser 3D visualization (via
react-three-fiber) that mirrors the physical state of your assets. - Fleet Orchestration: Manages the hierarchy of Workspaces, Projects, and Environments. It handles mission dispatch, ensuring that tasks defined in high-level JSON/YAML are routed to the correct edge agent.
- Edge Runtime: A deterministic execution environment running on the robot. Available as Python or C++ agents, it interfaces directly with hardware drivers (like ROS2) to execute local control loops while maintaining a connection to the cloud.
- Observability Pipeline: A high-throughput data layer built on MQTT for telemetry (joint states, battery, position) and WebRTC for low-latency video streaming. It ensures operators see what the robot sees, with sub-second latency.
- Data & ML: Dataset/episode capture for mission logs, plus ML model registry and training runs that let you iterate on policies with traceability.
How modules compose
These modules are loosely coupled but highly cohesive:
- Ingest: You upload an asset (URDF) to the Digital Twin module.
- Connect: You use the Edge Runtime (SDK) to connect a physical device, which authenticates and links to the Twin.
- Observe: The Edge Runtime streams data to the Observability Pipeline, animating the Twin in the dashboard.
- Control: You define a mission via Fleet Orchestration, which the Edge Runtime pulls and executes.