Investigation on the evolution of a robotic controller for autonomous vehicle navigation

Abstract

This work addresses the evolution of an Artificial Neural Network (ANN) to assist in the problem of autonomous navigation of a vehicle in urban environments. We propose a system architecture based on the use of two ANNs, one is trained for image processing, in charge of road recognition and employing template matching. The other ANN is evolved to perform the navigation control. The paper focuses on the evolved ANN, which provides steering and speed control to the autonomous vehicle, corroborating with the Evolutionary Robotic research area and showing its practical viability. The proposed system was tested in several experiments, evaluating not only the impact of different evolutionary computation parameters but also the role of the transfer functions on the evolution of the ANN. Results show that slight variations in the parameters lead to huge differences on the accuracy of the evolutionary process.

Increasing Intelligence in Inter-Vehicle Communications to Reduce Traffic Congestions: Experiments in Urban and Highway Environments

“…The type of ANN employed in our model is the Multi-Layer Perceptron, which is designed to find out the congestion levels in urban environments. The choice of ANN that used in our model is justified by: (i) does not need, every dissemination of data, changing the network topology [ 28 ]; (ii) generalize congestion detection in urban and highway environments; and (iii) has promising results in other areas of research [ 29 , 30 ]. ANN was configured with the following topology/features ( Fig 2 ): (i) two neurons at the input layer, representing the vehicle speed and density of neighboring vehicles; (ii) a hidden layer with four neurons, representing the learning ability, which can classify the congestion levels, it should be noted that four neurons were sufficient to obtain a high learning rate, as presented in [ 31 ]; and (iii) an output layer neuron representing the classification of the level of congestion on the roads.…”

Speed: Uma solução inter-veicular para detectar e controlar vias congestionadas no TMS

“…Diversas soluções foram propostas para lidar com o problema de congestionamento nosúltimos anos. Além de trazer maior conforto e segurança para os condutores [Pessin et al 2014, as soluções propostas tem como objetivo otimizar o tráfego de veículos no sistema de transporte, reduzindo o tempo de viagem, o consumo de combustível e a quantidade de CO 2 emitidos na atmosfera. Entretanto, modelar um mecanismo para estimar o nível de congestionamento da via com um aumento nos acertos das vias congestionadas, e com um tempo aceitável para alertar os condutores nãoé uma tarefa trivial.…”