Authors

Abstract

In this study, an autonomous navigation robot for poultry house inspection was designed, and a path optimization and obstacle avoidance strategy was proposed. First, a filtering algorithm was used to extract regions of interest from the 3D point cloud data collected by the inspection robot in caged poultry houses. Then, the geometric structure of cage-row lines was estimated using the least-squares method and refined using the RANSAC algorithm. The refined lines were projected to obtain boundary contour features. Finally, the A* algorithm was improved by removing redundant nodes, reducing the number of turning points, shortening the total path length, and increasing the weight of the cost estimation. The improved A* algorithm was also validated through physical robot simulation tests. Experimental results showed that compared with the least-squares method (LSM), the RANSAC-based approach achieved cage-row line slope values of 0.223 and 0.224 under Gaussian noise and manually added noise, respectively, demonstrating superior noise robustness and real-time performance. The results further indicate that the improved A* algorithm enhances path planning efficiency, enabling the robot to make timely decisions when encountering static or dynamic obstacles, thereby improving overall stability and reliability.

Abstract in Chinese

本研究设计了一种用于巡检禽舍的自主导航机器人，并提出了一种路径优化与避障策略。首先，利用滤波算法在巡检机器人采集的笼养禽舍三维点云数据中提取感兴趣区域；其次，利用最小二乘法（LSM）和随机抽样一致性算法（RANSAC）提取鸡舍线条并进行投影，获取边界轮廓特征；最后，对A*算法进行增强，消除冗余节点，减少路径转折点数量，缩短路径长度，增加代价估算的权重。同时，对改进的A*算法进行了实车仿真测试。实验结果表明，与LEM算法相比，RANSAC算法提取的鸡舍线条在高斯噪声和人工噪声下的斜率分别为0.223和0.224，RANSAC算法在抗噪能力和实时性方面均具有更优的表现。结果表明,改进的A*算法可以提高路径规划的效率,机器人在检测到动、静态障碍物时能够及时做出决策,具有更好的稳定性和可靠性。