目前，煤矿掘进机受巷道非结构化环境及复杂的工艺流程制约，自主化、智能化水平相对较低。特别在非全断面巷道作业时，主要依靠人工操作将掘进机移机进行二次截割完成断面成形任务，但由于井下环境恶劣，人工移机操作难度大、效率低下，且移机过程难以精准操控导致断面成形质量难以保证。因此，结合移动机器人轨迹规划技术、轨迹跟踪控制技术及断面成形截割控制技术，提出了煤矿巷道成形轨迹规划与悬臂式掘进机跟踪控制方法，通过分析掘进工作面特征和非全断面巷道成形作业流程，分别研究了井下非结构化环境下的掘进机移机轨迹规划和轨迹跟踪控制方法，实现掘进机的自主移机控制，在此基础上研究了断面二次成形轨迹规划及自动截割控制方法，最终实现了非全断面巷道成形截割控制，有效提升了巷道掘进的智能化程度和掘进效率，为实现掘进工作面少人化、无人化提供理论基础。

针对煤矿非全断面巷道条件下掘进机移机难度大、效率低下等问题，分析了煤矿井下非结构化环境特征及掘进机运动特性，提出了煤矿井下掘进机移机轨迹规划方法。利用深度相机将巷道环境实时重建，在虚拟环境中建立掘进机与巷道环境的碰撞检测模型，采用层次包围盒法进行碰撞检测，形成巷道边界受限下的避障策略。考虑到掘进机形体大小且路径规划过程目标单一，在传统SAC算法基础上，提出了HER-SAC算法，基于奖惩机制建立智能体，定义其状态空间与动作空间，并对智能体进行训练，得到掘进机的初始路径信息。根据初始路径结果进行速度规划，通过五次多项式插值法求解轨迹光滑函数，为掘进机的自主移机控制提供期望位姿。

针对井下环境复杂多变，掘进机的运动过程存在多种约束限制，导致移机控制过程困难、易受外界环境干扰且稳定性差等问题，提出了基于模型预测控制的掘进机移机轨迹跟踪控制方法。通过分析掘进机运动过程受巷道环境因素约束，采用牛顿-欧拉法建立了掘进机动力学模型，并选取状态变量和控制变量将其转化为状态空间方程。将动力学状态空间方程线性化处理后采用向前欧拉法近似离散化，得到控制器预测模型，设计控制器的代价函数及约束条件，再利用Lyapunov函数进行稳定分析，并设计两种参考轨迹进行仿真测试，验证了所提方法的有效性，从而实现了掘进机的移机轨迹跟踪控制。

针对非全断面巷道条件下掘进机单次截割无法满足巷道宽度要求，且人工操作二次截割过程效率低下，受巷道复杂环境影响难以精准操控，导致断面成形质量不佳等问题，提出了断面成形轨迹规划与自动截割控制方法。根据巷道设计参数和掘进机移机初始位置信息，制定了断面成形策略，设计了断面二次成形截割轨迹，采用D-H参数法进行掘进机截割臂运动学分析，求解各关节旋转变量，进而获得各关节运动轨迹。利用拉格朗日法建立了截割臂动力学模型，设计了基于反演滑模的自动截割控制器，并仿真验证了控制器在外界周期性扰动和随机扰动下均具有较强的鲁棒性，可实现对断面参考轨迹的精确跟踪。

最后，搭建实验平台，采用虚实结合的方式分别对掘进机移机轨迹规划功能、掘进机移机轨迹跟踪控制功能及断面二次截割控制功能进行测试验证。实验结果表明：所提轨迹规划方法在不同目标位置场景下均能够为机器人规划出平滑安全的移机可行轨迹；掘进机移机轨迹跟踪控制器能够稳定跟踪规划的移机参考轨迹，且跟踪效果良好；在移机初始位置和目标位置处的断面二次成形自动截割控制精度满足巷道断面成形质量要求。研究结果为煤矿井下掘进设备的自动控制提供了理论基础，为掘进工作面智能化发展提供了新方法。

At present, coal mine tunneling robots are restricted by the unstructured environment of roadways and complex technological processes, and the level of autonomy and intelligence is relatively low. Especially in the operation of the non-full-section roadway, the tunneling robot is mainly moved to the secondary cutting to complete the section forming task by manual operation, but due to the harsh underground environment, the manual relocation operation is difficult and the efficiency is low. Moreover, it is difficult to accurately control the moving process, which leads to the difficulty of ensuring the quality of section forming. Therefore, combined with the trajectory planning technology, trajectory tracking control technology and section forming and cutting control technology of mobile robot, a coal mine roadway forming trajectory planning and cantilever roadheader tracking control method was proposed. On this basis, the trajectory planning and automatic cutting control methods of the secondary forming trajectory of the section were studied, and finally the forming and cutting control of the non-full section roadway was realized, which effectively improved the intelligence and tunneling efficiency of the roadway, and provided a theoretical basis for the realization of less manpower and unmanned excavation of the excavation face.

Aiming at the problems of difficulty and low efficiency of roadheader moving under the condition of non-full section roadway in coal mine, the characteristics of unstructured environment in coal mine and the motion characteristics of roadheader robot are analyzed, and the trajectory planning method of robot moving in coal mine is proposed. The depth camera is used to reconstruct the roadway environment in real time, and the collision detection model between the tunneling robot and the roadway environment is established in the virtual environment. The hierarchical bounding box method is used for collision detection, and the obstacle avoidance strategy under the restricted roadway boundary is formed. Considering the size of the boring robot and the single goal of the path planning process, the HER-SAC algorithm is proposed on the basis of the traditional SAC algorithm. Based on the reward and punishment mechanism, the agent is established, its state space and action space are defined, and the agent is trained to obtain the initial path information of the boring robot. According to the initial path results, the velocity planning is carried out, and the trajectory smoothing function is solved by the quintic polynomial interpolation method, which provides the desired pose for the autonomous moving control of the boring robot.

Aiming at the complex and changeable underground environment, there are various constraints in the movement process of the tunneling robot, which lead to the difficulty of the movement control process, easy to be disturbed by the external environment and poor stability. A model predictive control based tracking control method for the movement trajectory of the tunneling robot was proposed. Based on the analysis that the movement process of boring machine is constrained by the roadway environment factors, the dynamic model of boring robot is established by Newton-Euler method, and the state variables and control variables are selected to transform it into the state space equation. After linearizing the dynamic state space equation, the forward Euler method is used to approximate discretization, and the controller prediction model is obtained. The cost function and constraint conditions of the controller are designed. The Lyapunov function is used for stability analysis, and two reference trajectes are designed for simulation test, which verifies the effectiveness of the proposed method. Thus, the moving trajectory tracking control of the boring robot is realized.

Aiming at the problems that the single cutting of the roadheader can not meet the width requirements of the roadway under the condition of non-full section, and the efficiency of manual operation of the secondary cutting process is low, and it is difficult to accurately control due to the complex environment of the roadway, resulting in poor section forming quality, a method of section forming trajectory planning and automatic cutting control is proposed. According to the tunnel design parameters and the initial position information of the TBM, the section forming strategy was formulated and the section cutting trajectory was designed. The D-H parameter method was used to carry out the kinematics analysis of TBM cutting arm, solve the rotation variables of each joint, and then obtain the motion trajectory of each joint. The dynamic model of the cutting arm is established by Lagrange method, and the automatic cutting controller based on inverse sliding mode is designed. Simulation results show that the controller has strong robustness under external periodic disturbance and random disturbance, and can accurately track the reference trajectory of the section.

Finally, an experimental platform was built to test and verify the trajectory planning function of roadheader moving, the trajectory tracking control function of roadheader moving and the secondary cutting control function of section respectively by using the method of combination of virtual and real. The experimental results show that the proposed trajectory planning method can plan smooth and safe moving trajectories for the robot under different target location scenarios. The moving track tracking controller can track the planned moving reference track stably, and the tracking effect is good. The automatic cutting control precision of section secondary forming at the initial position and the target position can meet the quality requirements of roadway section forming. The research results provide a theoretical basis for the automatic control of mining equipment and a new method for the intelligent development of mining face.