工程车辆快速智能化时的定位与三维环境地图创建研究

Engineering vehicles are widely used in the fields of construction and transportation. In the manual driving process, the operator needs high capability of hand-eye coordination. And it is difficult to be driven accurately and safely. This project will combine the mobile robotics and sensory technique, and reconstruct the engineering vehicle with fast automation and intelligentization.. In order to realize automatic exploration and navigation in natural environments, this project will mainly study the accurate localization for engineering vehicle as well as simultanenous localization and mapping in unknown environment. Firstly,since the motion model of the vehicle is unknown, the on-board laser range finder is used as the main exteroceptive sensor. After feature extraction and data association, an accurate localization method with smooth trace is proposed, so the steady control is guaranteed. Secondly, according to the three-dimensional scanning of the laser scanner, 3D point clouds data matching method is studied to acquire the volumetric map of the natural environment. Moreover, by incorporating the idea of submap and optimization of constraints, the globally consistent 3D mapping method is proposed to represent large scale environment.. This project uses fast reconstruction of intelligentization to endow the engineering vehicle with capabilities of localization, exploration and automatic navigation. In addition, it tries to lay the foundation of the application and industrialization of intelligent engineering vehicles on the aspects of theory and experiment.

工程车辆广泛应用于工程运输等领域。在人工驾驶时，对操作者的手眼协调能力要求较高，难以保证工程车辆精确、安全行驶。本项目将结合移动机器人和传感器技术，对工程车辆进行快速自动化和智能化改造。.为了实现在自然环境中自动探测与导航，本项目主要对工程车辆高精度定位、未知三维环境中的同时定位与建图方法展开研究。首先，针对工程车辆无运动模型的特点，使用车载激光测距传感器，通过特征提取和数据关联算法，建立具有平滑轨迹的精确定位方法，从而提高车辆行驶的平稳性。其次，通过激光雷达三维在线扫描，研究可靠的三维点云数据匹配方法，解决自然环境立体地图创建问题；并结合子地图思想和约束最优化方法，研究具有全局一致性的三维环境地图创建方法。.本项目通过快速智能化改造，使大功率液压工程车辆具有高精度定位与复杂环境探测能力，并为推广智能工程车辆的应用和实现产业化奠定理论和实验基础。

工程车辆在进行人工操作时，具有劳动强度大、控制精度低、过多依赖操作者经验等缺点。在进行智能化改造时，为了检测工程车辆的相对位姿，设计了一种增加全向轮轮组的方法，可以测量移动平台的运动参数和位姿变化。通过结合激光雷达等外部传感器，可以对工程车辆进行高精度定位和导航，达到厘米级精度。为了模拟多种工程车辆的运动方式，分别设计了两款全向移动机器人平台，分别使用麦克纳姆轮、中心可转向轮，可以用于室内、外环境的定位、探测和导航研究。未来将在智能移动平台上增加多自由度轻质操作臂，进行工业服务机器人研究，实现工程作业、智能搬运、装配和人机协作。