Authors

Abstract

This study proposes Smooth-Planner, a robust gradient-based local trajectory planning method for non-holonomic orchard robots operating in narrow row environments. The method employs a B-spline–based optimization framework that explicitly considers kinematic feasibility and collision avoidance without relying on dense global maps. Field experiments conducted in a standardized orchard demonstrate stable and repeatable performance, with lateral deviations effectively constrained and a minimum safety clearance of 0.90 m consistently maintained from crop rows. The results validate the practical safety and reliability of the proposed approach for autonomous agricultural operations.

Abstract in Chinese

本文提出了一种面向果园非完整约束移动机器人的稳健局部轨迹规划方法——Smooth-Planner。该方法基于 B 样条轨迹表示与梯度优化框架，在不依赖高精度全局地图的条件下，显式约束运动学可行性与避障安全性。标准化果园实地试验结果表明，该方法运行稳定、重复性良好，横向偏差得到有效控制，并始终保持不小于 0.90 m 的作物行安全间距。实验结果验证了该方法在自主农业作业中的安全性与实用性。