An Improved Theta* Algorithm for Mobile Robot Path Planning in Complex Environments

This paper proposes an improved Theta* path planning algorithm for mobile robot navigation in complex obstacle environments. The proposed method introduces a density-aware edge cost function that penalizes transitions into regions of high obstacle density while keeping the heuristic as the provably admissible Euclidean distance, improving both path smoothness and obstacle clearance. A direction-aware local obstacle density measure is designed to evaluate both the spatial concentration of obstacles and their angular orientation relative to the goal direction. A geometric weighting strategy imposes heavier penalties on obstacles located along the forward path toward the target, steering the search toward open regions without compromising the any-angle path capability inherent to Theta*. A line-of-sight sampling mechanism is introduced to accurately evaluate density for long-range jumps, and a precomputed density field reduces the per-query look-up cost. Simulation results demonstrate that the proposed method consistently achieves the fewest turning points and the lowest cumulative angle change and curvature among all compared algorithms, while maintaining competitive path lengths.