【摘要】 本研究面向半封闭牛场环境中的推料机器人导航需求，构建了A~*-TEB融合路径规划算法。针对传统A~*算法轨迹平滑度不足、节点冗余及障碍穿透风险，创新设计动态邻域切换机制：当节点扩展方向存在障碍时激活4邻域模式，无障碍威胁时切换8邻域模式，结合路径关键点筛选策略消除中间冗余节点，使路径长度缩短2.62%、规划效率提升3.18%、转折点下降34.85%。通过TEB(Time Elastic Band)算法二次优化构建时空弹性带，实现纵向偏差（4.5～39.5 cm，均值16.1 cm）与航向偏差（0.8°～4.7°，均值2.2°）的可控调节。速度提升阶段机器人位姿偏移量呈正相关增长，但全局精度波动幅度保持在预设阈值区间，当转向机构完成目标角度锁定时，通过多传感器融合的姿态补偿机制实现精度渐进收敛。融合TEB局部轨迹优化模块后实现局部路径修正，达到动态避障的目的，验证了算法的安全性和有效性。更多还原

【Abstract】 This study addresses the navigation requirements of feed-pushing robots in semi-enclosed cattle farm environments by developing an integrated A~* and Time Elastic Band(TEB) path planning algorithm. To overcome limitations of traditional A~*—including insufficient trajectory smoothness, redundant nodes, and obstacle penetration risks—a dynamic neighborhood switching mechanism is innovatively designed: when node expansion encounters obstacles, the algorithm switches to a 4-neighborhood mode; in obstacle-free conditions, it transitions to an 8-neighborhood mode. Combined with a key-point filtering strategy, redundant intermediate nodes are eliminated, achieving a 2.62% reduction in path length, a 3.18% improvement in planning efficiency, and a 34.85% decrease in turning points. The TEB algorithm is then employed for secondary optimization, constructing a spatio-temporal elastic band to achieve controllable regulation of longitudinal deviation(4.5–39.5 cm, mean 16.1 cm) and heading deviation(0.8°–4.7°, mean 2.2°). During speed-up phases, robot pose deviations exhibit positive correlation with speed growth, but global accuracy fluctuations remain within preset thresholds. When the steering mechanism locks the target angle, a multi-sensor fusion pose compensation mechanism ensures progressive convergence of precision. Integration of the TEB local trajectory optimization module enables dynamic obstacle avoidance through local path correction, validating the algorithm’s safety and effectiveness.更多还原