A path planning decision system for unknown pipeline detection using UAVs

This paper introduces the Pipeline-Aimed path Planner (PAPlanner), a sampling-based path planner for UAVs in unknown oil/gas pipelines. Its key contributions include a dynamic anchor point update strategy and path optimization that adapts to bends and diameter changes, eliminating redundant backtracking and enabling continuous exploration. By integrating real-time voxel maps, the algorithm optimizes paths to stay near the pipeline axis. Simulation results show that PAPlanner reduces average path length by 26.4% compared to the advanced MBPlanner method in the elbow scene experiment, demonstrating efficient safe trajectory maintenance. In the variable diameter scene experiment, where MBPlanner fails frequently, PAPlanner achieves a 0% failure rate. Real flight experiments validate its robustness with a 0.309 m average trajectory deviation from the axis, confirming reliable navigation. This work presents a novel framework enhancing UAV exploration efficiency in pipelines, overcoming limitations of existing algorithms for autonomous inspection in sensor-degraded confined environments.