基于双目视觉的掘进机器人定位定向方法研究

Localization and orientation method of roadheader robot based on binocular vision

随着煤矿采掘装备智能化需求日益迫切,煤矿掘进装备机器人化关键技术成为重要研究内容,确保掘进机器人的高精度定位定向是实现掘进机器人高效自主运行的重要基础,然而,受到掘进作业空间地质条件复杂、作业环境恶劣等因素影响,使得掘进机器人精确定位定向较为困难。基于双目视觉感知技术的掘进机器人定位定向方法,通过双目视觉传感器获取巷道空间环境特征,基于最大类间方差法建立巷道空间环境图像特征分割处理模型;分析巷道空间图像特征,构建掘进机器人及巷道空间环境特征识别模型;基于巷道空间环境图像识别信息,建立掘进机器人与巷道空间之间的坐标关系模型,推导并解算出掘进机器人在巷道空间中的位姿信息。经过仿真及实验验证表明,在实验室模拟巷道空间环境中,解算出的掘进机器人定位定向参数精度较好,算法误差较小,空间定位定向模型合理,仿真计算可靠,提高了掘进机器人对巷道空间环境信息的获取与特征识别能力,为掘进机器人的定位定向技术提供理论基础。

With the increasingly urgent demand for intelligent coal mining equipment,the key technology of coal mine excavation equipment robotization has become an important research subject.To ensure the high-precision positioning and orientation of the roadheader robot is an important foundation to realize its efficient autonomous operation.Due to the complex geological conditions and harsh work environment,which is difficult for the roadheader robot to accurately locate and orient,the paper explores the positioning and orientation method of the roadheader robot based on binocular vision perception technology.The spatial environment characteristics of tunnel were obtained by binocular vision sensor.Based on the Otsu Algorithm,the tunnel image feature segmentation model was established.By analyzing the spatial image features,the recognition model of roadheader robot and of tunnel space environment was constructed.With the image recognition information of the tunnel space environment in view,the spatial coordinate relationship model between the roadheader robot and the tunnel was established.The position and pose information of the roadheader robot in the tunnel space was deduced and solved.Simulation and experimental results show that the accuracy of positioning and orientation was higher in the laboratory simulation tunnel space environment.The algorithm error was small.The spatial positioning and orientation model was reasonable,and the simulation calculation was reliable.It improves the ability of the roadheader robot to obtain the spatial environment information as well as the feature recognition ability,and provides the theoretical basis for the positioning and orientation technology of the roadheader robot.