Controlling Computer-Generated Lifeforms using Fuzzy State Machine
(Room 320B)
04 Dec 19
8:30 AM
-
9:00 AM
Tracks:
Full Schedule, Wednesday Schedule
Simulated real-world environments represent a powerful tool for learning and training. Intelligent Virtual Agents (IVAs) can participate with people in such environments to facilitate training tasks that would normally require additional human participants. An IVA is a computer-generated lifeform that is not only visually similar to a human but also exhibits human-like behavior. An IVA can be seen as a simulated system situated in a virtual environment and capable of autonomous actions to meet its design objectives. Typical usage of IVAs in Naval Mission Training is Search and Rescue (SAR), Helicopter In-Flight Refueling (HIFR), or Vertical Replenishment (VertREP). In these applications, each IVA is assigned with a specific role to accomplish pre-defined tasks.
A suitable Modeling & Simulation approach to simulate the behavior of an IVA is a Finite State Machine (FSM). However, using an FSM to model human behavior has a couple of drawbacks:
An FSM can become very complex, especially when the IVA is required to perform its role autonomously and respond to the surrounding environment in a timely fashion.
A simpler FSM to model the IVA’s behavior can appear predictable and repetitive.
This paper proposes a simulation approach to reduce the predictability of the behavior of an IVA while enhancing their autonomy during the training by using Artificial Intelligence (AI) which can be achieved through the use of a Fuzzy State Machine (FuSM). First, the fuzzy logic technique and its basic rules are introduced. Then, the paper will explain the development process and how this technique can be applied to efficiently control IVAs behavior. The effectiveness of the FuSM modeling approach is assessed through the implementation of a typical Naval Mission Training application. Preliminary simulation results indicate that it is possible to easily control the behavior of computer-generated lifeforms using the FuSM modeling technique.