Enhancing Autonomous Navigation in GNSS-Denied Environment: Obstacle Avoidance Observability-Based Path Planning for ASLAM

Abstract

Obstacle avoidance (OA) is necessary for any path planning in outdoor environment to prevent any collision with the obstacles in natural environment. In this paper, a quadrotor navigates using Active Simultaneously Localization and Mapping (ASLAM) in GNSS-denied outdoor environment. In ASLAM, the quadrotor path is defined using real-time Observability Based Path Planning (OBPP) method, autonomously. To prepare using of the OBPP in outdoor environment, it is necessary to add the ability of OA to it. So, the OA-OBPP method is introduced which defines the path based on terrestrial landmarks while preventing any collision with the obstacles. This approach is developed by redefining a dataset of in range landmarks while all of the landmark in the vicinity of the obstacles are removed from the in-range landmarks dataset. To evaluate the performance of the proposed method, simulations of the OA-OBPP algorithm are conducted for a simplified 6-Degree of Freedom (DOF) quadrotor using MATLAB. The simulations evaluate the efficiency, accuracy and robustness of the proposed method. Results across various scenarios show that the method effectively avoids collisions with obstacles while simultaneously determining a path to the goal. Additionally, a comparison of the position estimation RMSE with Monte Carlo PP highlights the accuracy of the OA-OBPP. The robustness of the method, tested with varying initial positions, demonstrates its success in real-time path planning (PP) from any starting point to the destination without collisions. The results confirm that the OA-OBPP enhances the robot's capability to perform real-time, autonomous, and robust path planning in outdoor environments, even in the absence of GNSS signals, through visual navigation.