Path Planning of Mobile Robots Under Uncertain Navigation Environments Using FCM Clustering ANFIS

In this paper an improved multiple adaptive neuro-fuzzy inference system (MANFIS) to solve the wheeled mobile robots path planning in unstructured environments is proposed. The fuzzy C-means (FCM) clustering method is used in ANFIS to decrease the input data size, which leads to predict the efficiency of the proposed robot path planning model. The FCM clustering method allow classifying the robot sensors extracted input data into clusters; each cluster has similar properties that assists to develop the correlation between data and as a result simplify the proposed model. The design MANFIS architecture takes both the advantages of artificial neural network which has self-learning ability and fuzzy system to describe the uncertain phenomena of the data. Finally, by combining cluster data and MANFIS an optimum velocity for left and right wheel of the robot is determined, which safely navigate the robot in an optimized route. The simulation and experimental results show that the proposed path planning method is effective and can be implemented for any complex environments. The results obtained by the simulation and in physical experiments are equated with each other and noticed a good promise between both the results as the difference in results is less than 9%. The simulation results consistently demonstrate the superiority of the proposed FCM-ANFIS, manifesting performance improvements of up to 4.4% for path length reduction and up to 8.9% for mobile robot time duration reduction when compared to heuristic methods. This paper identifies and describes a new development on path planning technique that will help the robots to navigate in any kind of uncertain navigation environments.