Human Agents

The first step to creating the human agents within this software environment is to set up the data structure of the agent. As such, we need to create the following assets within the content browser: AI Controller, Blackboard, Behavior Tree, and Character Class.[1]

AI controller: This is the blueprint controller that controls the agent’s mental actions, which acts as the container for all decision branches the agent will utilize. From this, the AI controller will select a resulting state according to the agent’s choice and set it within the blackboard.

Blackboard: This is a container to store all the different states the agent can be in.

Behavior tree: This is a tool that controls all the physical actions of the agents. It will command a series of pre-established actions depending on the state of the agent set within the blackboard.

Character class: This blueprint class acts as a container for the physical attributes of the agents and is the physical entity that moves within the game environment. As such, this class would generally include assets such as the 3D model of the agent as well the textures that may be associated with it.

With this, we can establish a rough idea of how the agents will function within the software. As the agents explore the space, they will receive environmental input. They will then use this input to decide what action to take within the AI controller. This choice will then be used to set the agent state within the blackboard which will then be utilized by the behavior tree to determine a set of tasks the agent will perform. Once the agent finishes performing this task, it can then do additional tasks or return to a default state, in which the cycle will start over. Within the simulation, each agent will have their own independent copy of these assets, as such, they will function and interact independently from each other—much like how real humans do.

From here, the next step would be to re-establish the human systems from Chapter 2.3, which are the sensory system, the decision logic, and the pathfinding. We already established the components of this system, however, we can revisit our human systems from the top down instead of bottom up. This allows us to approach this problem in a sequential logical order to create our agents. There are, of course, various tools within UE4 to help us recreate these systems; therefore, this chapter will investigate the processes of doing so.