Intelligent Trajectory Planning and Navigational Analysis of Wheeled Mobile Robot in Cluttered Workspace

Abstract

Path planning and navigation analysis of wheeled mobile robots have received a significant role in the field of robotics by researchers in the past few decades. The research proposed by the researchers previously related to mobile robotics mainly considered three aspects such as localization, map building and path-planning. These aspects played an important role in finding out the feasible navigational path and it is also responsible for smooth or safe navigation in an environment. By keeping all these aspects in mind related to the solution of the trajectory planning problems, the trajectory planning and navigation control algorithms are presented in this proposed work. The proposed algorithms (BNN, DAYINDI AI, AGSA-DAYINDI AI and PSO-DAYINDI AI) can learn from the search space and configure themselves with the help of sensor modules. Based on this principle, the robot generates a collision-free path by avoiding obstacles from the source to the target. The primary objective of this research work is to design and develop smart computational intelligence techniques that addresses the online navigation problems as well as solves the problems using its learning feature. In this work, the path planning problems are addressed for unknown and messy environments using developed AI techniques. Two individual computational intelligence methodologies have been developed based on the Behaviour Based Neural Network (BNN) and DAYINDI AI algorithm. Also, hybrid methodologies have been developed by integrating the AGSA with DAYINDI AI and PSO with DAYINDI AI, to solve the mobile robot navigation problems. The performances of the techniques are examined individually, through simulation and real-time experiments. During the comparative study, good agreements (average deviation is less than 6%) have been found between simulation and real-time experiments. It has been noticed that AGSA-DAYINDI (deviation is less than 6%) and PSO-DAYINDI (deviation is less than 5.5%) hybrid controllers execute better results as compared to BNN (deviation is less than 6%) and DAYINDI AI (deviation is less than 6.5%) controllers. The navigational results obtained from the developed techniques are validated by comparing with the results of existing navigational techniques such as Fuzzy logic, Neuro-Fuzzy, Behavior-based Fuzzy, Heuristic approach, Heterogeneous ACO and SACOdm techniques. In comparison studies, it is found that the proposed methodologies generate better navigational results as compared to above existing techniques.