Precise path planning based on improved A* and DWA algorithms for complex agricultural environments

• Propose a Grey Wolf Optimizer-based bidirectional adaptive weighted A* algorithm. • Propose an adaptive Q-learning-DWA integrated architecture. • Establish a two-tier synergistic optimization scheme for global–local coordination. • Achieved precise navigation in agricultural environments. This study introduces a hybrid algorithm combining an enhanced A* algorithm with the Dynamic Window Approach (DWA) to improve agricultural robot navigation. The A* algorithm incorporates a Grey Wolf Optimizer (GWO) for dynamic heuristic weighting based on obstacle density and a bidirectional search strategy to boost efficiency. Path smoothing is achieved via key point selection and third-order B-spline fitting. In DWA, a Q-Learning mechanism adaptively optimizes weight coefficients, and a posture adjustment function eliminates initial heading deviations to avoid redundant steering. Simulations reveal that in simple environments, the proposed method reduces the computation time by 19.1% and increases the robot speed by 24.2%, with only a 1.7% increase in path length. In complex settings, it reduces the execution time by 20.1 %, shortens the path length by 1.2 %, and raises the speed by 25 %. Tests in a greenhouse demonstrate effective navigation in single- and multi-row operations, with 19.3 % and 37.7 % lower distance deviations, and 8.7 % and 14.6 % lower heading deviations, respectively. The experimental results demonstrate that the fusion algorithm, by simultaneously optimizing global and local path planning, significantly enhances the navigation accuracy, operational efficiency, and environmental adaptability of agricultural robots, thereby meeting the critical requirements for intelligent navigation in precision agriculture.