Remote agent: Difference between revisions

This is a set of three tasks revolving around the concept of a 'remote agent.' The overarching theme of the task set is to have a simple, stateful world with an agent that operates on the world's state until the end conditions are met. The world is a grid of colored squares, and is populated to some percentage by colored balls. The ending condition is when all balls are in a square of the correct color. The perspective of the agent is limited; it is not given an overview of the world. It only knows its surroundings by responses to its actions. Read on below for a complete description of the virtual world, interface commands and stream protocol.

 

# [[Remote agent/Agent interface|Implement the agent side of the stream interface in a manner idiomatic to the language]].

# [[Remote agent/Agent logic|Use the agent stream interface to satisfy the end conditions]].

 

Here is a description of the world and commands. The stream protocol below should be common among all implementations; any agent interface implementation should work equally well with any of the world simulation implementations.

* While the distribution and location of the color properties is undefined, there must not be more balls of a color than there are sectors with that color property.

* There must be at least one 'space' sector without a ball in it. (The agent does not initialize while holding a ball.)

 

== Goal ==

 

= Stream Protocol =

 

== Greeting ==

== Commands and event codes ==

 

|-

|name||type||character||decimal||hex

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|-

|}