A Perception-Decision-Control Framework for Dynamic Obstacle Avoidance of UAVs
Chen Chen · Advances in engineering research/Advances in Engineering Research · 2025
Unmanned aerial vehicles (UAVs) face significant challenges in dynamic obstacle avoidance despite their broad applications in agriculture, logistics, and emergency response.Existing methods, such as sampling-based RRT and graph-based Dijkstra algorithms, often struggle with dynamic environments and complex multi-objective tasks.This study proposes an integrated "perception-decision-control" framework to address these limitations.The framework incorporates multi-sensor fusion (LiDAR, vision, IMU) for robust environmental perception, linear temporal logic (LTL) for task decomposition and constraint formalization, and an improved Dijkstra algorithm with safety constraints for real-time path planning.A PID-based tracking controller ensures precise trajectory execution, achieving a 95% success rate in dynamic obstacle avoidance scenarios, with trajectory errors below 0.30.3 meters.Experimental validation on a quadrotor platform demonstrates enhanced adaptability to illumination variations and computational efficiency (45 FPS on embedded hardware).The system provides a comprehensive solution for autonomous UAV navigation, balancing safety, real-time performance, and mission complexity.Future work will focus on adaptive control strategies and multi-agent coordination to further improve robustness in large-scale environments.