Self-Recovery Algorithm for Autonomous Mobile Robots in Industrial Environments

This study presents a behavior tree-based approach for autonomous robot recovery in stuck situations, integrated with the ROS 2 and Navigation2 (Nav2) framework. The proposed system continuously monitors the robot’s environment using Lidar sensor data, partitioning it into angular sectors to detect potential immobilization. Upon identification of a stuck state, the robot initiates a multi-step recovery procedure. Initially, modified Nav2 parameters are applied to optimize navigation, followed by an escape maneuver that identifies feasible regions in the surrounding environment and guides the robot to a safe location. After successful recovery, the parameters are reverted to their default values to restore standard movement constraints. The implementation demonstrates a structured and modular approach to autonomous recovery, ensuring robustness and adaptability in dynamic environments. The methodology is validated through sequential execution of behavior tree nodes, each responsible for sensing, decision-making, parameter adjustment, and motion execution, highlighting the effectiveness of combining behavior trees with Nav2 for real-time robotic navigation