Multiagent mobility and lifestyle recommender system for individuals with visual impairment

Abstract

Background: Individuals with visual impairment currently rely on walking sticks and guide dogs for mobility. However, both tools require the user to have a mental map of the area and cannot help the user establish detailed information about their surroundings, including weather, location, and businesses.

Purpose and Methods: This study designed a navigation and recommendation system with context awareness for individuals with visual impairment. The study used Process for Agent Societies Specification and Implementation (PASSI), which is a multiagent development methodology that follows the Foundation for Intelligent Physical Agents framework. The model used the Agent Unified Modeling Language (AUML).

Results: The developed system contains a context awareness module and a multiagent system. The context awareness module collects data on user context through sensors and constructs a user profile. The user profile is transferred to the multiagent system for service recommendations. The multiagent system has four agents: a consultant agent, search agent, combination agent, and dispatch agent and integrates machine and deep learning. AUML tools were used to describe the implementation and structure of the system through use-case graphics and kit, sequence, class, and status diagrams.

Conclusions: The developed system understands the needs of the user through the context awareness module and finds services that best meet the user's needs through the agent recommendation mechanism. The system can be used on Android phones and tablets and improves the ease with which individuals with visual impairment can obtain the services they need.