Rover Sim : plataforma de validação de algoritmos aplicados a veículos autônomos para desvio de obstáculos.

Abstract

The development of autonomous vehicles is a contemporary theme, which aims to provide facilities for drivers and minimize traffic problems generated by them. Many of the research in the area aim at solving problems related to vehicle locomotion in complex environments that can contain static and dynamic obstacles. This paper presents the development of a validation platform to obstacle avoidance algorithms, applied to autonomous vehicles. The goal is to help define the best obstacle avoidance strategy for a given environment, and assist in the definition of sensors to be used. For the validations were implemented the Tangent Bug and Artificial Potential Fields obstacle avoidance algorithms, in addition to the implementation of simulation of ultrasonic and laser sensors. The tests performed had as objective to verify the behavior of the two algorithms in different environments, in order to identify, from parameters such as the number of collisions, targets reached and displacement performed, which of the algorithms performed the defined path presenting the best results. The Tangent Bug algorithm was able to perform all defined paths in the environments in which it acted, while the Artificial Potential Fields algorithm, in one of the environments, was not able to perform the course. As for the sensors, the laser sensors presented superior results to the ultrasonic sensors when the Tangent Bug algorithm was used. In the use of the algorithm Artificial Potential Fields, both sensors demonstrated a similar performance.