Positioning and Navigation Technology of Farm Inspection Robot Based on Improved Dynamic Window and A* Algorithm

As a key terminal of smart agriculture, the positioning and navigation technology of farm inspection robots directly affects the continuity and reliability of inspection operations. To achieve millimeter level positioning and efficient path planning of robots in farm environments, this study dynamically adjusts the weight of the A* algorithm heuristic function by quantifying obstacle density, and introduces a coincidence evaluation function in the dynamic window to enhance the algorithm's predictive ability. The improved correlation scanning and matching positioning method using artificial landmarks and multi-resolution maps enhances the stability of positioning in complex environments. The outcomes demonstrate that the improved algorithm has a total path length of 48.470m and a total turning angle of 351.78 ° in a 30 × 30 grid map, which is 6.4% shorter than the conventional algorithm. The improved positioning algorithm has an average error of 1.998mm and 2.417mm in the X and Y directions under 5 landmark observations, with a heading angle error of 0.335 °. It maintains an accuracy of 4mm even when the landmarks are obstructed, and successfully avoids dynamic obstacles in actual farm environments. This technology effectively improves the navigation robustness and positioning accuracy of farm inspection robots in complex environments through collaborative optimization of path planning and positioning methods, providing technical support for autonomous inspection of smart agriculture.