Path finding for autonomous vehicles by Orientation-Aware Advanced: A* Algorithm in a grid map

This paper presents an Orientation-Aware Advanced A* algorithm (OA-A*) designed to incorporate heading changes into path planning for robots operating under a stop–turn–go motion model. The method introduces a tunable weighting parameter that regulates the trade-off between geometric path length and turning effort. Extensive simulations were conducted on three representative grid environments with varying structural complexity to evaluate path length, turning behavior, smoothness, computational effort, and total travel time. The results indicate that, under specific parameter settings and map conditions, OA-A* can achieve reductions of up to approximately 60% in directional changes relative to baseline grid-based planning while maintaining comparable path lengths. Across the evaluated scenarios, the method consistently demonstrates controllable trade-off behavior, enabling smoother trajectories when turning costs are emphasized. Real-world experiments using an automated electric wheelchair platform equipped with a 2D LiDAR sensor validate the practical applicability of the approach and confirm trends observed in simulation. The experiments show that increasing the orientation weighting reduces reorientation frequency and can improve traversal efficiency under physical execution constraints. Additionally, comparisons between Euclidean and Octile heuristics indicate similar path structures with differences primarily in computational performance. Overall, the findings suggest that incorporating orientation awareness into grid-based planning enhances flexibility and supports adaptation to robot-specific motion characteristics while remaining computationally feasible for structured environments.