本文为加拿大康科迪亚大学（作者：MohammadAli AskariHemmat）的硕士论文，共118页。

本文以差分驱动机器人为研究对象，提出了多个不同路径规划算法的发展与实现。在此基础上，提出了一种实时路径规划器，能够在非结构环境中为非完整无人车（UGV）找到最优的无碰撞路径。首先，设计并实现了一个混合A*路径规划器来寻找最优路径，实时将无人值守地面车辆的当前位置与目标相连，同时避开无人值守地面车辆附近的任何障碍物。该路径规划器的优点是，利用势场技术，排除了障碍物周围的节点，大大减少了传统A*方法的搜索空间；它还能够根据障碍物的性质和安全考虑给予它们不同的优先次序，从而区分不同类型的障碍物。这种方法对于保证在人类与自主车辆密切接触的环境中安全驾驶至关重要。然后，考虑UGV的运动学约束，生成光滑的、可驾驶的几何路径。在整个论文中，我们进行了大量的实际实验，以验证所提出的路径规划方法的有效性。

This thesis work proposes the development and implementation of multiple different path planning algorithms for autonomous mobile robots, with a focus on differentially driven robots. Then, it continues to propose a real-time path planner that is capable of finding the optimal, collision-free path for a nonholonomic Unmanned Ground Vehicle (UGV) in an unstructured environment. First, a hybrid A* path planner is designed and implemented to find the optimal path; connecting the current position of the UGV to the target in real-time while avoiding any obstacles in the vicinity of the UGV. The advantages of this path planner are that, using the potential field techniques and by excluding the nodes surrounding every obstacles, it significantly reduces the search space of the traditional A* approach; it is also capable of distinguishing different types of obstacles by giving them distinct priorities based on their natures and safety concerns. Such an approach is essential to guarantee a safe navigation in the environment where humans are in close contact with autonomous vehicles. Then, with consideration of the kinematic constraints of the UGV, a smooth and drivable geometric path is generated. Throughout the whole thesis, extensive practical experiments are conducted to verify the effectiveness of the proposed path planning methodologies.

1.  引言

2.  数学建模

3.  势场函数

4.  基于启发式的路径规划

5.  具体实现与结果

6.  结论与未来工作展望

更多精彩文章请关注公众号：