Automatic navigation of wall-following mobile robot using a hybrid metaheuristic assisted neural network

This work proposes an attempt to control the navigation of a wall following mobile robot using a hybrid metaheuristic which is a combination of two different population-based heuristics such as Gravitational Search (GS) and Particle Swarm Optimization (PSO). The hybrid metaheuristic is called GSPSO which is used to train a multi-layered Artificial Neural Network (ANN) with the available benchmark sensor readings datasets obtained from the navigation of a wall following mobile robot SCITOS G5. The performance of the proposed model was evaluated with regard to different tuning parameters present in the metaheuristic as well as in the ANN. Simulated experimental results show satisfactory performance which makes the proposed method adoptable in practice for controlling a mobile robot for predicting its next direction based on sensed distance information from the wall or obstacles surrounding it.