Obstacle Avoidance System Using Artificial Neural Network and Fail-Safe PLC

Abstract

Autonomous vehicles constitute the next revolution in transportation. They are designed to improve road efficiency and driving safety. Autonomous vehicles are capable, using artificial intelligence, of recognizing their surroundings, navigating, and avoiding obstacles without human intervention. Integrating Artificial Neural Networks (ANNs) into the safety Programmable Logic Controllers (fail-safe PLCs) to create an algorithm that controls an autonomous vehicle and ensures safety on the roads is presented. ANNs are a supervised machine-learning model based on a computing system built to simulate the way the human brain processes and analyzes information. A fail-safe PLC offers a safety concept in the field of machine and personnel protection. The fail-safe system is connected directly to the ANN’s program. It is responsible for shutting down the ANN in cases where the ANN output starts providing false results that could danger humans or property. Also, the fail-safe PLC works in conjunction with the main PLC to serve as a backup to the neural network system. A set of training examples involving over 30 data was evaluated to train the ANN. In addition, a fail-safe PLC program was designed to perform under special conditions. Indoor obstacle avoidance courses were used as an example to examine the effectiveness of the obstacle avoidance system. Simulation results show that the system successfully predicted and responded correctly to the training data and avoided all obstacles on the obstacle-avoidance courses. The system examined real-world data and the experimental results verify the effectiveness of the proposed method.