This research aims to enhance the traditional A* algorithm to improve path planning performance in robotics by focusing on generating smoother, shorter and more efficient paths. The key objectives include minimizing path length, reducing sharp turns and producing paths suitable for real-world robotic navigation.
The proposed method integrates a grid-based environment model with a modified A* algorithm, followed by post-processing techniques such as B-spline smoothing, line-of-sight shortcutting and gradient descent optimization. The process involves six main steps: environment modeling, heuristic adjustment, path finding, smoothing, shortcutting and performance tuning. Performance is evaluated based on metrics such as path length, computation time, and success rate using a simulation-based assessment model.
Simulation testing demonstrated that the enhanced algorithm achieved an average path length of 26.4 steps, a computation time of 15.2 ms, and a 100% success rate, indicating significant improvements in efficiency and reliability.
This study contributes a hybrid enhancement to the A* algorithm that balances computational efficiency and path quality through a combination of classical and optimization-based techniques. It sets the foundation for future work involving machine learning for adaptive path optimization in dynamic, real-time robotic environments.
