<p>&nbsp; &nbsp; 煤矿智能化建设对于煤矿安全、高效、绿色生产具有重要意义。煤矿巷道顶板支护工艺流程复杂，并且钻锚装备自动化智能化程度低，导致煤矿巷道支护速度慢，跟不上巷道掘进的速度，巷道&ldquo;掘支失衡&rdquo;问题长期存在，提高钻锚装备自动化智能化水平是提高巷道支护效率，破解难题的关键。因此，本文提出了锚固孔视觉定位与钻锚机器人自主控制方法，利用视觉检测方法实现锚固孔快速识别与精确定位，构建机械臂视觉伺服控制模型，控制机械臂运动实现末端执行器快速精确对准目标锚固孔中心。通过提高钻锚装备自动化智能化程度，实现了提高巷道顶板支护速度的目标。</p> <p>&nbsp; &nbsp; 针对人工识别定位锚固孔存在的位置偏差大、劳动强度大和安全性差等问题，提出基于参数自适应调整Hough变换（Hough Transform）的锚固孔自动识别检测方法。通过构建成像平面上锚固孔尺寸预测模型，基于钻锚机器人运动学模型自适应调整Hough变换参数，实现锚固孔快速识别检测；利用锚固孔物理信息构建几何约束，筛除无效锚固孔识别检测结果，提高锚固孔识别检测精度；基于轮廓形状和几何约束完成图像立体匹配，构建双目视觉定位模型完成锚固孔空间位置解算，实现煤矿巷道锚固孔快速识别与精确定位，为后续实现钻锚机器人自主控制奠定基础。</p> <p>针对不符合Pieper准则的机械臂逆运动学求解存在的收敛难和精度差等问题，提出基于最优初始值与迭代步长因子的机械臂逆运动学求解方法。首先构建描述机械臂末端执行器空间位姿的误差函数，基于蒙特卡洛随机采样方法分析机械臂工作空间，利用空间点与工作空间映射关系确定误差函数最小值对应的空间点，将其对应的关节角度和位置作为迭代方法求机械臂逆解的最优初始值，解决了迭代方法随机初始值引起的难以收敛问题；针对迭代过程固定步长导致的收敛时间长或振荡问题，引入迭代步长因子，构建实时动态自适应调整策略，解决了迭代方法容易陷入局部最优的难题，提高了钻锚机器人逆运动学求解精度和速度。</p> <p>&nbsp; &nbsp; 人工控制钻锚装备完成支护任务存在精度低、安全性差等难题，针对所述问题提出基于分步视觉伺服的钻锚机器人运动控制方法。在锚固孔快速识别与精确定位的基础上，利用五次多项式插值方法进行机械臂轨迹规划，构建基于位置的视觉控制模型，控制机械臂按照规划轨迹运动，实现末端执行器接近目标锚固孔位置，完成机械臂粗控制；当末端执行器与目标位置距离小于设定阈值时，以目标锚固孔外接矩形顶点坐标为图像特征，构建基于图像的视觉伺服控制模型，利用粒子群优化算法进行图像雅可比矩阵估计，设计系统的运动控制率，实现末端执行器快速精确对准目标锚固孔中心，解决了人工控制钻锚装备导致的控制精度低和安全性差的难题。</p> <p>&nbsp; &nbsp; 为验证锚固孔视觉定位与钻锚机器人自主控制方法的有效性和精确性，搭建钻锚机器人实验平台，进行锚固孔视觉识别检测与定位实验和钻锚机器人视觉伺服控制实验。实验结果表明，本文所述锚固孔视觉识别检测与定位方法精度高、实时性好，能够实现锚固孔快速识别与精确定位；分步视觉伺服控制方法能够控制机械臂运动，实现末端执行器快速精确对准目标锚固孔中心。本文提出的锚固孔视觉定位与钻锚机器人自主控制方法有效提高了煤矿钻锚装备自动化程度，提升了巷道顶板支护效率，有效缓解了&ldquo;掘支失衡&rdquo;难题，为综掘工作面智能化建设提供一种新思路。</p>

<p>&nbsp; &nbsp; The development of intelligent systems in coal mines is critically important for ensuring safe, efficient, and environmentally sustainable production. The process of supporting the roofs of mine tunnels is complex, and the automation and intelligence levels of anchor drilling equipment are low. This results in slow support installation, unable to keep pace with tunnel excavation, leading to a persistent imbalance between excavation and support. Enhancing the automation and intelligence of anchor drilling equipment is key to improving support efficiency and addressing this challenge. Therefore, this study proposes a method for visual positioning of drilling holes and autonomous control of anchor drilling robots. By utilizing visual detection methods, the rapid identification and precise positioning of drilling holes can be achieved. A visual servo control model for the manipulator is constructed to control its movement, enabling the end effector to quickly and accurately align with the center of the target drilling hole. By enhancing the automation and intelligence of anchor drilling equipment, the goal of accelerating the support process for mine tunnel roofs is realized.</p> <p>To address the issues of large positional deviations, high labor intensity, and poor safety associated with manual identification and positioning of drilling holes, an automatic detection method for drilling holes based on parameter adaptive adjustment of the Hough Transform is proposed. By constructing a prediction model for drilling hole dimensions on the imaging plane and adaptively adjusting the Hough Transform parameters based on the kinematic model of the anchor drilling robot, rapid identification and detection of drilling holes are achieved. Geometric constraints are constructed using the physical information of the drilling holes to filter out invalid detection results, thereby improving detection accuracy. Contour shape and geometric constraints are used for image stereo matching, constructing a binocular vision positioning model to calculate the spatial position of the drilling holes. This enables rapid identification and precise positioning of drilling holes in coal mine tunnels, laying the foundation for the subsequent autonomous control of the anchor drilling robot.</p> <p>To address the issues of convergence difficulty and low accuracy in solving the inverse kinematics of manipulators that do not conform to Pieper&#39;s criterion, a method for solving the inverse kinematics of manipulators based on optimal initial values and dynamic iterative step coefficient is proposed. First, an error function describing the spatial pose of the manipulator&#39;s end effector is constructed. The Monte Carlo random sampling method is used to analyze the manipulator&#39;s workspace, and the mapping relationship between spatial points and the workspace is utilized to determine the spatial point corresponding to the minimum error function value. The joint angles and positions corresponding to this point are taken as the optimal initial values for the iterative method to solve the inverse kinematics, addressing the convergence issues caused by random initial values in iterative methods. To tackle the problems of long convergence time or oscillation caused by a fixed step size in the iterative process, an iterative step coefficient is introduced to construct a real-time dynamic adaptive adjustment strategy. This approach mitigates the risk of the iterative method getting trapped in local optima, thereby enhancing the accuracy and speed of solving the inverse kinematics for the manipulator of the anchor drilling robot.</p> <p>Manual control of anchor drilling equipment for support tasks faces challenges such as low precision and poor safety. To address these issues, a stepwise visual servo control method for the manipulator of an anchor drilling robot is proposed. Based on the rapid identification and precise positioning of drilling holes, the manipulator&#39;s trajectory is planned using a quintic polynomial interpolation method. A position-based visual control model is constructed to control the manipulator&#39;s movement along the planned trajectory, allowing the end effector to approach the target drilling hole position, thus achieving coarse control of the manipulator. When the distance between the end effector and the target position is less than a set threshold, the coordinates of the vertices of the bounding rectangle around the target drilling hole are used as image features. An image-based visual servo control model is then constructed. The particle swarm optimization (PSO) algorithm is employed to estimate the image Jacobian matrix, and the system&#39;s motion control rate is designed to enable the end effector to quickly and accurately align with the center of the target drilling hole. This approach resolves the issues of low control precision and poor safety associated with manual control of anchor drilling equipment.</p> <p>To validate the effectiveness and accuracy of the visual positioning and autonomous control methods for the anchor drilling robot, an experimental platform was constructed. Experiments were conducted on the visual identification and positioning of drilling holes as well as the visual servo control of the manipulator of the anchor drilling robot. The experimental results demonstrated that the proposed method for visual identification and positioning of drilling holes is highly accurate and offers good real-time performance, enabling rapid and precise identification and positioning of drilling holes. The stepwise visual servo control method effectively controls the manipulator&#39;s movement, allowing the end effector to quickly and accurately align with the center of the target drilling hole. The proposed methods for visual positioning and autonomous control of the anchor drilling robot significantly enhance the automation of coal mine anchor drilling equipment, improve the efficiency of tunnel roof support, and effectively mitigate the issue of &quot;excavation-support imbalance.&quot; This provides a novel approach for the intelligent development of comprehensive tunneling operations.</p>