Skripsi
INTEGRASI DAN EVALUASI ALGORITMA HYBRID A-STAR DAN DYNAMIC WINDOW APPROACH UNTUK NAVIGASI CERDAS AUTONOMOUS VEHICLE
The development of autonomous vehicles requires a reliable navigation system to determine optimal routes and avoid obstacles in real-time, especially in semi-structural environments such as campus areas. This study aims to integrate and evaluate the Hybrid A* algorithm as a global planner and the Dynamic Window Approach (DWA) as a local planner. The research methodology involved simulation testing using CARLA software on the Town01 map with a Nissan Micra vehicle model, as well as real-time route data collection using a Toyota Raize car at the Universitas Sriwijaya environment for trajectory analysis. Simulation results demonstrated that the integration of both algorithms successfully navigated the vehicle with a 100% success rate, where the DWA algorithm was able to maintain a minimum safety distance of over 3 meters from static and dynamic obstacles11. Meanwhile, the evaluation of real-world route data indicated that the Hybrid A* algorithm produced kinematically valid trajectories but resulted in longer travel distances compared to Google Maps and Real Drive references. This was caused by the algorithm's response to GPS noise, which generated corrective maneuvers in a sawtooth pattern to comply with the vehicle's non-holonomic constraints. This study concludes that the integration of Hybrid A* and DWA is effective in prioritizing safety and generating drivable paths consistent with vehicle dynamics. Keyword: Autonomous Vehicle, Hybrid A*, Dynamic Window Approach (DWA), Path Planning, Intelligent Navigation.
No other version available