Mobile Robot Localization in Indoor Environments Using a Hybrid Extended Kalman Filter and Rapidly-Exploring Random Tree Algorithm

Abstract

This paper presents an integrated framework for mobile robot navigation in indoor environments, based on the fusion of the Extended Kalman Filter (EKF) and the Rapidly-exploring Random Tree* (RRT) algorithm. The objective of this hybrid approach is to simultaneously enhance path planning accuracy and localization precision under conditions of sensor noise and dynamic, uncertain obstacles. Initially, the RRT algorithm generates an optimal path from the start to the goal point, considering existing obstacles. Subsequently, for high-accuracy path following, the robot's position is estimated in real-time using the EKF. In this method, the prediction step of the filter relies on the robot's kinematic model and odometry data, while the update step is performed using Lidar sensor measurements. The precise localization provided by the EKF enables the robot to follow the proposed RRT* path with minimal error. A key feature of this framework is the closed-loop interaction between localization and path planning; if the estimation error increases or unexpected environmental changes occur, the path planning process is reactivated by the RRT* module. This design increases the system's adaptability and stability in the face of environmental uncertainties. Simulation results demonstrate that this framework significantly improves localization accuracy, achieving a Root Mean Square Error (RMSE) of less than 0.055 and a 73% increase in precision compared to pure odometry. Furthermore, the paths generated by RRT* show considerable improvement in length, smoothness, and safety over those produced by the basic RRT algorithm.