Authors

Abstract

Addressing issues such as slow path planning speed, high path costs, and visual positioning errors encountered by grain harvesters during field operations, this study proposes an improved rapidly-exploring random trees with visual servoing (VS-IRRT) algorithm. By leveraging visual servoing technology to real-time acquire environmental information, the algorithm enables precise positioning and attitude correction of the harvester; Based on this, heuristic sampling strategies and path optimization functions are introduced to enhance the node expansion efficiency and convergence speed of the random rapid search tree. To reduce path costs, a path evaluation model based on environmental feature costs is designed, comprehensively considering factors such as terrain complexity, crop distribution density, and machine turning radius, dynamically adjusting the search direction and optimizing path smoothness. Simulation and field test results show that the VS-IRRT algorithm reduces path planning time by approximately 32% compared to traditional RRT, reduces average yaw error by 42%, reduces path curvature change rate by 33%, and reduces turning frequency by 21%. It also maintains high robustness and planning accuracy even in the presence of visual noise and positioning errors. This study provides an effective path planning method and technical support for autonomous navigation and efficient operation of grain harvesters in complex agricultural environments.

Abstract in Chinese

针对谷物收割机在田间工作过程中存在的路径规划速度慢、路径成本高以及视觉定位误差等问题，提出结合视觉伺服的改进随机快速搜索树算法 (Improved rapidly-exploring random trees with visual servoing, VS-IRRT)过视觉伺服技术实时获取作业环境信息，实现收割机的精准定位与姿态修正；在此基础上，引入启发式采样策略和路径优化函数，提高随机快速搜索树的节点扩展效率与收敛速度。为降低路径成本，设计基于环境特征代价的路径评估模型，综合考虑地形复杂度、作物分布密度及机器转向半径等因素，动态调整搜索方向并优化路径平滑性。仿真与田间试验结果表明，VS-IRRT算法在路径规划时间上较传统RRT缩短约32%，平均偏航误差降低42%，路径曲率变化率降低33%，转向次数减少21%。且在存在视觉噪声与定位偏差的情况下仍能保持较高的鲁棒性与规划精度。该研究为谷物收割机在复杂农田环境下的自主导航与高效作业提供了有效的路径规划方法与技术支撑。