Actors

Actors are self-contained entities such as AMRs, humans using digital interfaces, AI agents, and so on. They have persistent state and skills, capable of autonomous decision-making and interaction within the Paranet’s ecosystem. All actors are written in Paraflow, though their skills may be implemented in other languages such as JavaScript or Python.

Actors are the intelligence of the network based on the skills that are published to their nodes for discovery by other actors. The nodes are actors, but this is from a systems viewpoint, not for developers. Developers build actors that are either entirely Paraflow based or are in an application language and actorized. To actorize is to map functionality to skills for which there are mapping translators provided (currently only for Python in this version of the devkit).

A Python-based robot running in the Omniverse (NVIDIA’s VR platform) will publish skills to its node as a Paraflow program automatically. Once translated, there will exist both the python robot code and the actor Paraflow code that enables that robot to collaborate on the Paranet. The primary functions are in Python and access to the network is handled by a Paraflow wrapper. An orchestrator might be written in Paraflow to manage the operations of the Omniverse environment.

Actors are models written in Paraflow that describe their skills, events, rules, and tasks. An “actor” is written Paraflow source code for developers. System actors may be written in Paraflow or Rust (not available to devkit users).

Actor Instances are Paraflow code running in the Paraflow interpreter with state. An instance may be live or dormant. Events engage the state machine of the instance. Dormancy can be short or long running—there is no limit to the dormancy length of these lambda-like instance machines. If power is lost the instance machine will recover to its last state.

The following is a conceptual view of actor models, not an instance representation. Actors can exist anywhere but are shown here within a conceptual container of a node. The Paranet can be considered as a network of skills. Nodes broker PnCP requests to match the best skill at runtime. Actors implement skills.

Diagram D2

Actors A1-A7 are shown as part of their respective nodes with their skills.