2.3.2. Acoustic

Masking auditory signals by binaurally re-displaying environmental information through headphones or earbuds blocks vital environmental signals on which many visually impaired people rely for secure navigation [22]. Currently, bone-transmitting helmets enable the user to obtain 3D auditory feedback, leaving the ear canal open and enabling the user to operate with free eyes, hands, and mind. The algorithm reduces sound production that does not indicate a change in order to further minimize the auditory output. Thus, the audible output sound is only produced when the user is confronted with an impediment, limiting possibly irritating sounds to a minimum [85].

### *2.4. Functions and Applications*

Here, we review the necessary functions and applications that the visually impaired use to solve difficult matters. These applications are obstacle detection, navigation, facial recognition, color and texture recognition, micro-interactions, text recognition, informing services, and braille displayers and printers. Next, a detailed review is presented.

### 2.4.1. Obstacle Detection

A grea<sup>t</sup> deal of research and many studies have focused on obstacle detection due to its significance for the visually impaired, as it is considered to be a major challenge for them. An ETA using a microwave radar to detect obstacles up to the head level through the vertical beam of the sensor was presented in [37]. To overcome the issue of power consumption, the authors switched off the transmitter during the listening time of the echo. Moreover, the pulsed chirp scheme was adapted to manage the spatial resolution. To improve the precision of the indoor blind guide robot's obstacle recognition, Du et al. [86] presented a sensor data fusion approach based on the DS evidence theory of the genetic algorithm. The system uses ultrasonic sensors, infrared sensors, and LiDAR to collect data from the

surroundings. The optimized weight is replaced in DS evidence theory by data fusion for the purpose of determining the weight range of various sensors using the genetic algorithm. In practice, weighing and fusing evidence requires the determination of the weight of the evidence. Their technique has an accuracy of 0.94 for indoor obstacle identification. Bleau et al. [87] presented EyeCane, which can identify four kinds of obstacles: cubes, doors, posts, and steps. However, its bottom sensor failed to properly identify objects on the ground, making downwards navigation more dangerous.
