A Novel Path Planning and Obstacle Avoidance Method for Considerable Localizing Error

Mobile robots hold significant potential for a broad spectrum of applications in industry and various service domains. Consequently, extensive research efforts have been devoted to addressing deficiencies and improving their performance. Among the critical challenges in robotics is obstacle avoidance, which enables the robot to navigate around unexpected objects encountered along a planned path. Numerous methods and algorithms have been proposed to prevent collisions between robots and detected obstacles. These approaches commonly rely on the crucial assumption of having precise knowledge of the robot's position at every step. This paper introduces a novel method for obstacle avoidance in indoor environments, leveraging an occupancy grid map of a partially known space and the A* algorithm. The proposed method addresses scenarios with imprecise information about the robot's state. Initially, a preliminary occupancy grid map is refined and transformed into an enhanced map using an artificial neural network. Subsequently, the A* algorithm is applied to the modified map. Additionally, an algorithm is developed to guide the robot from a starting point to a target endpoint. When encountering a newly emerged obstacle, the robot dynamically adapts its path to reach the goal while avoiding the obstacle. The proposed method's efficacy is validated through simulations of a two-wheeled robot in three distinct scenarios. Results demonstrate the method’s capability to navigate the robot effectively within an indoor environment, even with imprecise state information. The algorithm ensures the robot maintains a safe distance from obstacles, showcasing its potential for practical applications.