UGV Navigation in Complex Environment: An Approach Integrating Security Detection and Obstacle Avoidance Control

To address the challenges faced by Unmanned Ground Vehicles (UGV) in autonomously navigating complex environments, particularly in scenarios with high security requirements in topological road map (TR-Map) scenes, our research integrates the environmental adaptability of Partially Observable Markov Decision Processes (POMDP) with the Artificial Potential Field (APF) method. We propose a safety detection strategy and introduce road potential fields, developing a Model Predictive-based Potential Field Control (MP-PFC) algorithm. By deeply integrating obstacle avoidance mechanisms with global path tracking and leveraging POMDP to enhance the system's capability to handle environmental uncertainties, we establish a navigation framework that is both efficient and secure. The experimental scenarios in this paper include security detection, trap escape, complex scene control, control performance at different speeds, as well as outdoor comprehensive and interactive tests, thoroughly validating the system's safety and adaptability. The test results show that the proposed approach can not only enhance the security and adaptability of UGV in complex environments, but also provide a robust solution for the autonomous navigation of unmanned systems.