煤矿井下发生事故后，救援机器人已经成为实施灾难援救工作的主要设备之一，但多数的救援机器人仍依靠于救援人员的远程控制，其实时性和精确性与操作人员的熟练度密切相关，机器人的自主性受到很大限制。因此，本文提出并构建的煤矿蛇形搜寻机器人系统，研究蛇形搜寻机器人在煤矿井下事故现场局部未知环境中的姿态控制与路径规划，目的是实现煤矿蛇形搜寻机器人系统的自主导航与控制，对于解决煤矿搜寻机器人在井下局部未知的环境中自主运动控制问题具有重要的意义。 针对本文提出并构建的蛇形搜寻机器人体系结构的设计问题，在对其功能分析的基础上对系统的整体方案进行设计，通过对比，确定机械材料的使用及处理器的类型，设计整个控制系统的硬件电路和软件程序，完成实验样机的制作。针对蛇形搜寻机器人的控制问题，以PC机作为上位机、STM32F407ZET6作为下位机的控制器，上位机主要起监控作用，通过无线通讯遥控蛇形搜寻机器人的运动方式并接收机器人采集到的各种传感信息。 针对蛇形搜寻机器人的姿态控制问题，通过对其运动关节机构的运动学分析，提出将多自由度高度冗余的蛇形机器人抽象为空间连杆结构，构建了两连杆之三关节的结构模型，并在此基础上建立了正交连接的蛇形搜寻机器人的D-H坐标系。针对在煤矿井下不确定路况下的蛇形搜寻机器人的姿态控制问题，研究了机器人应对多种不同路况的运动控制函数；设计并提出了基于改进Serpenoid曲线的煤矿蛇形搜寻机器人姿态行为控制方法，使其针对不同的路况条件能作出相应的姿态运动。仿真实验表明采用该姿态控制算法后机器人的越障过程稳定性明显提高。 针对煤矿井下环境的地图构建问题，结合激光雷达获取障碍信息的数据特点，提出一种基于栅格拓扑地图的局部未知障碍物几何特征地图匹配的方法（Geometric Map Matching Algorithm for Local Unknown Obstacles based on Grid Topology Map，GMA-LUO-GTM），该方法以栅格拓扑地图作为搜寻机器人的局部未知障碍物环境的地图模型；利用几何特征地图匹配的方法，寻找出障碍物与机器人之间相对的位形关系，并解析出机器人在当前位置的可行路径网络，通过激光雷达对多种障碍物的检测及实验表明，其位姿横纵坐标及朝向角（ ）的标准差分别为（1.45，1.01，0.1），表明该方法可实现对机器人的精确定位。 针对蛇形搜寻机器人的全局路径规划问题，提出基于优化人工势场栅格蚁群算法（Optimization of Artificial Potential Field Grid Ant Colony Algorithm，OAPF-GACA）的路径规划方法。首先依据矿井实际环境利用栅格生成地图模型，再采用优化的人工势场法对蚁群算法的信息素进行初始化，使其在起始时刻就有好的引导作用，算法在前期的收敛速度有了较大提升，并采用参数可自动调节的蚁周模型，避免了算法易陷入局部最小和过早停滞的状况。最后，对在栅格地图上产生的算法路径再优化，使路径的距离和转角得到明显的改善，更适合机器人去追随。实验证明，与传统的蚁群算法对比，其收敛迭代次数减少了50%，优化后的路径长度缩短了18.7%。基于OAPF-GACA的路径的规划方法可以在不同的障碍环境中实现机器人的全局的路径规划。 针对蛇形搜寻机器人的未知局部环境中的路径规划问题，基于对栅格地图的分析，提出了可插入多个局部目标点的改进人工势场算法（Improved Artificial Potential Field Algorithm for Inserting Multiple Local Target Points，IAPF-IMLTP）。该算法在OAPF-GACA得到的全局优化路径的基础上，插入了多个不同的局部的目标点，每次迭代只有靠机器人最近且在其前方的局部的目标点起作用，再通过改进其引力势函数，可以使机器人贴合规划路径行进。实验表明，IAPF-IMLTP算法生成的路径的与规划路径相比平均误差只有4.9%。 针对环境中障碍物为动态的避障路径规划行为，提出了具有参数自适应调整的多目标点牵引的人工势场算法（Artificial Potential Field Algorithm for Multi Target Traction with Parameter Adaptive Adjustment，APF-MTT-PAA）的局部路径规划方法，该算法采用斥力参数自适应调整策略，并通过赋予障碍物一个虚拟速度的方法，消除了机器人易进入局部的最小值的状态。大量的实验仿真和实测实验表明，该算法可以使搜寻机器人在不同的障碍环境中迅速抵达目标。与传统的人工势场法相比，其避障速度快，可寻最优路径，无局部不可达的情况。 实验及测试表明，所研制的蛇形搜寻机器人具有不错的实时性和自主性，能够满足搜寻机器人在较为复杂的环境下的自主控制的需求。本文提出的路径规划算法可为煤矿搜寻机器人在未知障碍环境中的路径规划提供理论支撑，也对机器人在高危环境的导航策略研究奠定了基础。

After the coal mine accident, rescue robot has become one of the most important equipment for disaster relief work, but most of the rescue robots still rely on the remote control of rescue workers. The reality and accuracy are closely related to the proficiency of operators. The autonomy is very limited. Therefore, the system of snake-like searching robot in the coal mine is proposed, which study the attitude control and path planning of the snake-like searching robot in the local unknown environment of the mine accident site. The purpose is to realize the local autonomous navigation and autonomous control of the coal mine snake-like searching robot system. It is of great significance to solve the issue of autonomous movement control of coal mine search robots in a partially unknown underground environment. Aiming at the design problem of the architecture of snake-searching robot by the dissertation proposed, based on its function analysis the whole scheme of the system is designed. Through comparison, the use of mechanical materials and the type of processor are determined, and the entire control system is designed. The hardware circuit and software program complete the production of experimental prototypes. For the control problem of snake-like searching robot, the PC is used as the host computer and the STM32F407ZET6 is used as the controller of the lower computer. The host computer mainly plays a monitoring role, and through the wireless communication remote control snake-like searching robot's movement mode and receiving information of various sensors collected by the robot. Aiming at the attitude control problem of snake-like searching robot, a kinematics analysis of its kinematic joint mechanism is proposed to abstract the multi-degree-of-freedom highly redundant serpentine robot into a spatial link structure, and the model of the three joint structure of two connecting rod is constructed. On this basis, the D-H coordinates of the snake-like robot with orthogonal connection are established. Aiming at the attitude control problem of the snake-like searching robot under uncertain road conditions in the coal mine, the robot's motion control function to deal with a variety of different road conditions was studied, and the attitude behavior control method of the coal snake-like searching robot based on the improved Serpenoid curve was designed and proposed. It can make corresponding gesture movements for different road conditions. Simulation experiments show that the robot's obstacle-obstacle process stability is significantly improved by using this attitude control algorithm. Aiming at the problem of map construction in coal mine underground environment, combined with the data characteristics of laser radar acquisition obstacle information, a method of Geometric Map Matching Algorithm for Local Unknown Obstacles based on Grid Topology Map (GMA-LUO-GTM) was proposed, This method uses a grid topology map as a map model for searching the robot's local unknown obstacle environment. It uses a geometric feature map matching method to find the relative positional relationship between the obstacle and the robot, and resolves the robot's current position., and analyze the current situation of the robot in the feasible path through the network, laser radar detection and simulation experiments on a variety of obstacles show that this method can find the robot relative to the configuration relationship of obstacles. The detection and experimentation of various obstacles by Radar shows that the standard deviations of the pose's horizontal and vertical coordinates and orientation angle ( ) are respectively (1.45, 1.01, 0.1), indicating that this method can achieve accurate positioning of the robot. Aiming at the global path planning problem of snake-like searching robot, a path planning method based on Optimization of Artificial Potential Field Grid Ant Colony Algorithm (OAPF-GACA) is proposed. Firstly, the map is generated by raster based on the actual environment of the mine, and then the optimal artificial potential field method is used to set the initial pheromone of the ant colony algorithm, so that it has a good guiding role and improves the convergence speed of the algorithm in the early stage. The ant cycle model which can adjust the parameters automatically avoids the local optimization and the premature stagnation. Finally, the algorithm path generated on the grid map is re-optimized, so that the distance and the rotation angle of the path are significantly improved, which is more suitable for robots to follow. Experiments show that compared with the traditional ant colony algorithm, the number of convergence iterations is reduced by 50%, and the optimized path length is shortened by 18.7%. The path planning method based on OAPF-GACA can implement the global path planning of the robot in different obstacle environments. Aiming at the local unknown environment path planning problem of snake-like search robot, Based on the analysis of the grid map, an improved artificial potential field algorithm which can insert multiple local target points (IAPF-IMLTP) is proposed. Based on the global optimization path obtained by OAPF-GACA, by setting up a series of local target points, each iteration plays a role only with the nearest target and the local target point in front of it. Then, by improving its gravitational potential function, the robot can move forward according to the optimal path. Experiments show that the average error of the path generated by the IAPF-IMLTP algorithm is only 4.9% compared with the planned path. Aiming at obstacle avoidance path planning behavior of dynamic obstacles in the environment, a local path planning method of Artificial Potential Field Algorithm for Multi Target Traction with Parameter Adaptive Adjustment (APF-MTT-PAA) is proposed. The algorithm uses a self-adaptive adjustment strategy for repulsive forces and eliminates the robot's tendency to enter a local minimum by giving a virtual velocity to the obstacle. A large number of experimental simulations and measured experiments show that the algorithm can make the search robot reach the target quickly in different obstacle environments. Compared with the traditional artificial potential field method, the obstacle avoidance speed is fast, the optimal path can be found, and there is no local inaccessibility. Experiments and tests show that the snake-like searching robot has good real-time and autonomous ability, and can satisfy the requirement of autonomous control of searching robot in more complicated environment. The path planning algorithm proposed in this dissertation can provide theoretical support for the path planning of coal mine searching robots in unknown obstacle environments, and also lays a foundation for the research of robot navigation strategy in high-risk environment.