Swarm

简体中文

Domain Specific Language of Multi-Agent System Based on both Rules and Neural Networks

Design Principles

The cluster mission of unmanned aerial vehicles (UAVs) can be divided into three layers:

• The Task layer, responsible for coordinating and scheduling multiple UAVs for collaboration and information exchange.
• The Behavior layer, describing the action sequences of individual UAVs.
• The Action layer, which describes the execution process of an atomic action of a UAV. This process is controlled by rule-based or neural network-based syntax.

Lexcial Conventions

A program consists of one or more translation units stored in a file. The program is reduced to a sequence of tokens.

Tokens

• Identifiers
• Keywords
• Constants
• Operators

Identifiers

Identifiers begin with a letter or underscore and consist of letters, digits, or underscores. They are case-sensitive. Identifiers can represent various entities such as variables, task names, behavior names, action names, message topics, agent types, etc.

Keywords

• Task, Behavior, Action
• @subtask, @behavior, @action
• @uavTypes, @topic, @init, @goal, @routine
• publish, subscribe, request, from, to
• each, order
• if, else, return
• Agent

Constants

• Integer Constants, also known as decimal integers.
• Floating Constants, consisting of an integer part, a decimal point, and a fractional part.
• String Constants, a sequence of characters surrounded by double quotation quotes, such as "Hello, swarm."

Operators

Grammar

EBNF

Examples

To make examples in VSCode highlight as shown in the image, follow these steps:

• install plug-in Highlight
• modify the setting.json of your vscode as vscode-setting.json