Mapping and autonomous navigation of TurtleBot2 using hybridization of A* algorithm, DWA, and RatSLAM

Abstract AMR navigation in unknown and dynamic environments needs to be robust, efficient, and reliable. For this, integration of a RatSLAM-based localization and cognitive mapping with the A* global planner and Dynamic Window Approach (DWA) local planner in a ROS2 environment is a very efficient method. The novelty of the proposed framework is the unified integration of the biologically inspired RatSLAM localization system with the global and local planning systems. It enhances the consistency of navigation and the simulation with the real deployment of the system. The localization and the cognitive mapping information from RatSLAM are continuously provided for spatial awareness. A* computes the global path from the current position to the target, and DWA generates real-time collision-free velocity commands while satisfying the robot's kinematic and dynamic constraints. In this hierarchical interaction, localization, global planning, and local obstacle avoidance are coupled together for autonomous navigation. The framework is tested on the TurtleBot2 platform in both Gazebo simulation and the real world in various navigation scenarios. The experimental results show that navigation performance is reliable and operation remains stable even with changes in the environment. Path length reduction is around 2-3% when compared to individual navigation components and a standard A*-DWA navigation pipeline. Travel times are reduced by 7-11%. The results show that the combination of RatSLAM and both global and local planning modules increases the reliability of localization, stability of navigation and autonomous navigation performance.