Lugo4Node is a NodeJS implementation of a client player for Lugo game.
It is not a bot that plays the game, it is only the client to connect to the game server.
This package implements many methods that does not affect the player intelligence/behaviour/decisions. It is meant to reduce the developer concerns on communication, protocols, attributes, etc.
Using this client, you just need to implement the Artificial Intelligence of your player and some other few methods to support your strategy (see the example folder and The Dummies JS).
- Requirements
- Installation
- Usage
- First option: Implementing a Bot class (simpler and recommended)
- Second option: Implementing the turn handler (a little more work)
- Third option: Using reinforcement learning 🧠
- NPM >= 16
npm install @lugobots/lugo4node
There are three ways to use Lugo4Node client:
See example
Lugo4Node PlayAsBot implements a very basic logic to reduce the code boilerplate. This client will wrap most repetitive code that handles the raw data got by the bot and will identify the player state.
The Bot interface requires the methods to handle each player state based on the ball possession.
class Bot {
// OnDisputing is called when no one has the ball possession
onDisputing(inspector: GameSnapshotInspector): Order[] | { orders: Order[], debug_message: string } | null {
// the magic code comes here
return ...
}
// OnDefending is called when an opponent player has the ball possession
onDefending(inspector: GameSnapshotInspector): Order[] | { orders: Order[], debug_message: string } | null {
// the magic code comes here
return ...
}
// OnHolding is called when this bot has the ball possession
onHolding(inspector: GameSnapshotInspector): Order[] | { orders: Order[], debug_message: string } | null {
// the magic code comes here
return ...
}
// OnSupporting is called when a teammate player has the ball possession
onSupporting(inspector: GameSnapshotInspector): Order[] | { orders: Order[], debug_message: string } | null {
// the magic code comes here
return ...
}
// AsGoalkeeper is only called when this bot is the goalkeeper (number 1). This method is called on every turn,
// and the player state is passed at the last parameter.
asGoalkeeper(inspector: GameSnapshotInspector, state: PLAYER_STATE): Order[] | { orders: Order[], debug_message: string } | null {
// the magic code comes here
return ...
}
}As you noticed, the option 1 has some logic injected in it, so you may want to remove that layer of logic and implement yours.
The most raw way to communicate with the lugo is receiving the game raw snapshots. See example
In this case, you must use the client's play method. The play method will call your call back function for any
message received from the game lugo.
It may require that you know a bit more about the game steps, but still not too much.
If you are a machine learning enthusiastic you may want to use the Lugo reinforcement learning environment.
Lugo bots is an asynchronous game, so you will need to use the Lugo4Node Gym library to create your model:
See example and documentation at RL lib readme file
There are a many things that you will repeatedly need to do on your bot code, e.g. getting your bot position, creating a move/kick/catch order, finding your teammates positions, etc.
Lugo4Node brings some libraries to help you with that:
Auto generated doc coming soon
The Snapshot inspector is quite useful. Firs to it helps you to extract data from the Game Snapshot each game turn.
const inspector = new GameSnapshotInspector(proto.lugo.Team.Side.HOME, 8, snapshot);
inspector.getSnapshot(): Lugo.GameSnapshot | null
inspector.getTurn(): number
inspector.getMe(): Lugo.Player
inspector.getBall(): Lugo.Ball | null
inspector.getPlayer(side: Lugo.Team.Side, number: number): Lugo.Player | null
inspector.getBallHolder(): Lugo.Player | null
inspector.isBallHolder(player: Lugo.Player): boolean
inspector.getTeam(side: Lugo.Team.Side): Lugo.Team | null
inspector.getMyTeam(): Lugo.Team | null
inspector.getOpponentTeam(): Lugo.Team | null
inspector.getMyTeamSide(): Lugo.Team.Side
inspector.getOpponentSide(): Lugo.Team.Side
inspector.getMyTeamPlayers(): Lugo.Player[]
inspector.getOpponentPlayers(): Lugo.Player[]
inspector.getMyTeamGoalkeeper(): Lugo.Player | null
inspector.getOpponentGoalkeeper(): Lugo.Player | nullAnd also help us to create the Turn Orders Set based on the game state and our bot team side:
inspector.makeOrderMove(target: Lugo.Point, speed: number): Lugo.Order
inspector.makeOrderMoveMaxSpeed(target: Lugo.Point): Lugo.Order
inspector.makeOrderMoveFromPoint(origin: Lugo.Point, target: Lugo.Point, speed: number): Lugo.Order
inspector.makeOrderMoveFromVector(direction: Lugo.Vector, speed: number): Lugo.Order
inspector.makeOrderMoveByDirection(direction: DIRECTION, speed?: number): Lugo.Order
inspector.makeOrderMoveToStop(): Lugo.Order
inspector.makeOrderJump(target: Lugo.Point, speed: number): Lugo.Order
inspector.makeOrderKick(target: Lugo.Point, speed: number): Lugo.Order
inspector.makeOrderKickMaxSpeed(target: Lugo.Point): Lugo.Order
inspector.makeOrderCatch(): Lugo.OrderNaturally, the bots see the game field based on coordinates x and y, as in a cartesian plane.
However, that's not something that we want to be concerned about during the bot development.
The classes Mapper and Region work together to facilitate it for use.
Mapper slices the field in columns and rows, so your bot does not have to care about precise coordinates or the team
side. The mapper will automatically translate the map position to the bot side.
And you may define how many columns/rows your field will be divided into.
const map = new Mapper(10, 10, teamSide)
// you may find a Map Region based in coordinates:
const region = map.getRegion(2, 3)
// and you may find a Map Region based in a map point
const region = map.getRegionFromPoint(inspector.getPlayer(proto.lugo.Team.Side.AWAY, 5))
// and you can also know the position of the goals
map.getAttackGoal(): Goal
map.getDefenseGoal(): GoalThe Region helps your bot to see the game map in quadrants, so it can move over the field without caring about coordinates or team side.
region.front()
region.back()
region.left()
region.right()