A Framework for the Generation of Obstacle Data for the Study of Obstacle Detection by Ultrasonic Sensors
Bhupendra Singh, Monit Kapoor
Abstract
In addition to numerous problems faced by the Visually Impaired persons, navigation through the outdoor environment is one of them, as the terrain is unknown with surfaces filled with obstacles of different heights. Several Electronic Travel Aids are proposed in the literature, and the methodology is evaluated by placing different types of the obstacle, traveling through them blindfolded, and counting the number of times obstacles are detected or not by the system. In our view, this approach helps little for further advancing the current system as the experimental scenario is not repeatable entirely and makes it difficult to compare new research findings with any existing one. We therefore, present a model inspired by the real world settings for detecting low-lying obstacles, considering few assumptions and constraints. The simulations are also proposed based on the model to generate obstacle data any number of time. Furthermore, to validate our model and simulations, field experiments are conducted, and the results obtained are compared with the simulation result. The evaluation parameter in terms of standard deviation, accuracy, false positive, errors encountered suggest that the proposed model and simulations effectively encapsulates the real-world parameters and thus can be explored and utilized further by the research community.