FRB-BlinGui: A Fuzzy Rule-Based Model for Predicting Collision Risks with Obstacles in Assistance to Visually Impaired People.
People with visual impairment, Obstacle detection, Fuzzy rules, Time series, Fuzzy rule-based systems, Collision predictions, Collision risks.
It is possible to observe in the literature that several approaches are developed for safe navigation and orientation of people with visual impairments, aiming to avoid collisions with obstacles. These approaches often use ultrasonic or infrared sensors, mobile applications with cameras for computer vision, or wearable devices. However, uncertainties and inaccuracies resulting from the dynamics of the environment are often overlooked, making navigation and orientation for people with visual impairments insecure and subject to various types of collisions with different types of obstacles. Faced with this situation, there is a need to develop approaches that consider the degree of collision risk presented by each obstacle, the dynamics in the displacement between the individual and the obstacle, and its location at high or low points of complex perception. In this context, the fuzzy set theory (FST) presents itself as an essential tool for dealing with the imprecision and uncertainties existing in the environment since the use of systems based on FST has the advantage of intuitively interpreting user behavior. Thus, this thesis presents an innovative obstacle detection and collision risk prediction model called FRB-BlinGui in dynamic scenarios. The proposed model was tested on a wearable device to detect obstacles and prevent collisions in real time, offering gradual alerts about risks. The results show its effectiveness in providing alerts, promising to improve the safety and mobility of people with visual impairments.