Design of the Integrated Cognitive Perception Model for Developing Situation-Awareness of an Autonomous Smart Agent

Abstract

This study explores the potential for autonomous agents to develop environmental awareness through perceptual attention. The main objective is to design a perception system architecture that mimics human-like perception, enabling smart agents to establish effective communication with humans and their surroundings. Overcoming the challenges of modeling the agent's environment and addressing the coordination issues of multi-modal perceptual stimuli is crucial for achieving this goal. Existing research falls short in meeting these requirements, prompting the introduction of a novel solution: a cognitive multi-modal integrated perception system. This computational framework incorporates fundamental feature extraction, recognition tasks, and spatial-temporal inference while facilitating the modeling of perceptual attention and awareness. To evaluate its performance, experimental tests and verification are conducted using a software framework integrated into a sandbox game platform. The model's effectiveness is assessed through a simple interaction scenario. The study's results demonstrate the successful validation of the proposed research questions.