机场跑道混凝土摊铺车辆环境感知与轨迹规划系统设计  

Design of environment perception and trajectory planning system for concrete paving vehicles on airport runways

作　　者：吴霄峰 张逸飞 宋雅楠 卓海清 熊树生[1] WU Xiaofeng;ZHANG Yifei;SONG Yanan;ZHUO Haiqing;XIONG Shusheng

机构地区：[1]浙江大学,浙江杭州310012 [2]福州海卓马克机电设备有限公司,福建福州310100

出　　处：《现代机械》2024年第5期7-13,共7页Modern Machinery

摘　　要：在机场混凝土道面无人化施工的工程背景下,针对道面摊铺与后处理车辆所涉及关键技术进行研究,设计了一种基于多模态传感器的环境感知和轨迹规划系统。通过激光雷达和可见光相机,对采集到的混凝土道面边界信息进行处理和计算,最终规划出可引导摊铺机行驶的参考轨迹。环境感知模块采用了基于深度学习的图像检测技术和基于点云的特征提取手段,采用卡尔曼滤波算法进行多传感器数据融合;轨迹规划模块通过最小化冲击度的5次多项式差值法进行轨迹规划。实验结果表明,该系统能够实现对道面边界的准确提取,并规划出一条无碰撞且动力学可行的参考轨迹,能够满足车辆行驶与施工的需要,具有一定的应用价值与工程意义。Under the context of unmanned construction of airport concrete pavements,this paper focuses on the key technologies involved in pavement paving and post-processing vehicles.A multi-modal sensor-based environment perception and trajectory planning system is designed.Through the laser radar and visible light camera,the collected boundary information of the concrete pavement is processed and computed,generating a reference trajectory for guiding the paving machine.The environment perception module adopts deep learning-based image detection technique and point cloud-based feature extractiontechnique,combining multiple sensor data through the Kalman filter algorithm.The trajectory planning module utilizes a 5th order polynomial interpolation method that minimizes the impact force to plan the trajectory.Experimental results demonstrate that the system can accurately extract the pavement boundary and generate a collision-free and dynamically feasible reference trajectory,which can meet the requirements of vehicle travel and construction application,and has certain application value and engineering significance.

关 键 词：履带式车辆 环境感知 传感器融合 轨迹规划 

分 类 号：U461.1[机械工程—车辆工程]