无人驾驶装载机自主铲掘作业关键技术研究

The health and safety risks caused by whole body vibration and stability failure have been a long-standing problem for wheel loader operators when the equipment moving or working on unstructured terrain. Because of the potential benefits of relieving the operators from the complex driving and working conditions, the unmanned driving technology has generated considerable recent research interest. In the respect of operational process, this research focuses on the intelligent digging technology for wheel loaders dealing with bulk materials. To be more specific, the particle model calibrated by bulk material properties, the multi-body kinematics and dynamic model of wheel loader would be constructed at the beginning of the first stage of this research to analyze the working performance of wheel loader under different working conditions and different digging trajectories through the method of simulation. Then, based on the simulation results, the theoretical digging trajectory would be optimized for different kinds of material in different pilled-up condition. Furthermore, taking the time delay characteristic of the hydraulic system into consideration, the control method for digging trajectory tracking would be studied. In the second stage of this research, in view of the inconsiderable digging conditions, the behavioral mechanism of experienced operators should be taken into comprehensive considerations serving as the foundation for generating proper digging strategies. After all the theoretical calculation and analysis as describe above, in the final stage of this research, the autonomous digging control system would be designed, and the corresponding experimental device would be established. Field tests would be taken for realizing the research object of an efficient autonomous digging with low energy consumption and high numerical value of bucket fill factor. The achievements of this research will provide new research foundations and technical support for Autonomous operation of wheel loaders, which will be of great theoretical and practical significance in promoting the intelligent process of construction machinery in our country.

装载机在非结构路面上行驶和作业时振动剧烈、稳定性差，驾驶员承受着振动和失稳带来的健康和安全威胁，无人驾驶技术作为当前研究的热点，能将驾驶员从装载机复杂恶劣的行驶和作业环境中解脱出来。本项目从作业角度出发，研究装载机对散状物料的自主铲掘技术。首先建立包含物料、整机运动学和动力学特性的装载机仿真模型，分析不同作业工况和铲掘轨迹下的装载机作业性能，优选针对不同属性和堆放形态物料的理论铲掘轨迹，并开展考虑液压时滞特性的铲掘轨迹跟踪控制研究；综合考虑非理想铲掘工况下的驾驶员行为机制，制定基于仿驾驶员行为的铲掘策略；在此基础上，设计装载机自主铲掘作业控制系统并搭建硬件平台，开展相关试验研究，实现装载机在短时间内以低能耗获得高满斗率的自主作业目标。本课题的研究成果将为实现装载机无人驾驶提供新的研究基础和技术支持，对于推动我国工程机械的智能化进程具有重要的理论和现实意义。

无人驾驶装载机是矿山、港口和基建等领域智能化建设拼图中的核心装备，自主铲掘是实现其完全智能化作业的关键环节。本项目主要完成了以下工作：.首先，建立了装载机铲掘动力学模型，针对装载机在铲掘过程中铲掘力难以直接测量的问题，提出了一种基于无迹卡尔曼滤波的装载机铲掘力估计方法，铲装过程中的最大估计误差低于5%。.其次，针对装载机面向不同物料时需自适应调整铲掘策略的要求，提出了基于雷达图法的铲掘难易程度评估方法。确定了包含铲掘时间、时间变化率、油缸压力、压力变化率等信息的铲掘难易程度综合评估指标，并通过聚类算法和雷达图方法综合评估了堆积物料的铲掘难易程度。.然后，针对装载机铲掘作业性能主要评价指标“满斗率”的测量问题，提出了两种不同的满斗率测量方法。第一种方法是基于双目视觉三维重构及点云处理算法，应用了点云融合和表面插值等技术，即使在视野遮挡的情况下也能达到94.72%精度。第二种方法将双目摄像机捕获的图像用于生成点云，提出了一种用于点云深度学习的满斗率测量方法，平均测量精度为95.23%。.最后，提出了面向不同作业物料的装载机自主铲掘方法。通过分析装载机铲掘过程确定了铲掘过程参数及其铲掘性能评价指标，形成了基于Kriging模型和遗传算法的装载机铲掘过程优化方法，实现了以低能耗模式获得高满斗率的作业目标。总结实际铲掘过程中可能遇到的铲斗无法铲入到预定深度、动臂无法举升和车轮滑转等异常工况以及驾驶员应对策略，制定了基于仿驾驶员行为的装载机铲掘策略，保证了装载机在面对不同类型物料时都能够完成自主铲掘作业。.综上所述，本项目建立了装载机铲掘作业联仿模型并准确估计铲掘力大小，研究了物料铲掘难易程度评估方法和满斗率测量方法，提出了面向易铲物料的铲掘过程优化方法，并制定了面向难铲物料的自主铲掘策略。本项目的研究成果将为装载机自主作业提供新的研究基础和技术支持，对于推动我国工程机械的智能化进程具有重要的理论和现实意义。