【摘要】 针对传统波前边缘检测算法在探索过程中随已知区域的扩大导致提取探索候选点收敛时间上升、传统评价函数选取探索目标点时未考虑机器人姿态，使得未知环境下自主探索效率低的问题，提出一种融合局部与全局视角的自主探索算法。所提方法采用限制传统波前边缘检测（WFD）算法遍历范围，并与快速搜索随机树算法同步运行，分别提取局部与全局探索候选点；利用轻简化后的萤火虫算法接受探索候选点进行聚类；通过在具有信息增益、路径代价的传统评价函数中引入角度代价构成改进的评价函数实现最佳探索目标点的选取，引导机器人完成未知环境的高效探索；最后，基于机器人操作系统（ROS）搭建仿真与样机测试平台进行了实验验证，结果表明所提算法较基于快速随机搜索树-广度优先搜索（RRT-BFS）算法在探索耗时与探索路径层面分别缩减了22.19%和32.13%，提升了自主探索效率。更多还原

【Abstract】 Aiming at the problems that the convergence time of extracting exploration candidate points increases with the expansion of known regions in the exploration process of traditional wavefront edge detection algorithm, and the traditional evaluation function does not consider the robot pose when selecting exploration resulting in inefficient autonomous exploration in unknown environment, an autonomous exploration algorithm combining local and global perspectives is proposed. By limiting the traversal range of the traditional WFD algorithm and running synchronously with the fast search random tree algorithm, the proposed method extracts local and global exploration candidate points respectively, and uses the light simplified firefly algorithm to accept exploration candidate points for clustering. An improved evaluation function is formed by introducing the angular cost into the traditional evaluation function with information gain and path cost to realize the selection of the best exploration target points and guide the robot to complete the efficient exploration of unknown environment. Finally, a simulation and prototype experiment platform based on the robot operating system(ROS) is built for experimental validation, and the results showed that the proposed algorithm reduced the exploration time and exploration path level by 22.19% and 32.13%,respectively, compared with the RRT-BFS algorithm, which improves the efficiency of autonomous exploration.更多还原