基于几何规则和前向模拟的机器人自主探索

Robot Autonomous Exploration Based on Geometry Rules and Forward Simulation

在未知环境下进行自主探索和地图创建问题,现有自主探索方法主要集中于基于边界扫描的自主探索,其计算量大、效率低下、在复杂环境中耗时长。针对此问题,提出一种基于探索区域几何规则的前向模拟(G-FS)方法。该方法利用探索区域的几何规则、序贯蒙特卡洛规划选取最终目标点。该方法有效增加了机器人单次探索的移动距离,并解决了前向模拟方法的局部问题,提高了自主探索的效率。通过G-FS与前向模拟、边界扫描方法的对比实验证明,在相同环境下,所提出的G-FS算法计算量大幅降低,制图所耗时间明显减少,效率提高。

Aiming at the problem of autonomous exploration and map creation in unknown environment, the existing autonomous exploration methods mainly focus on autonomous exploration based on boundary scanning, which has large amount of calculation, low efficiency and long time-consuming in complex environment. To solve this problem, a forward simulation method based on exploring regional geometric rules is proposed, which is called geometry forward simulation(G-FS) method. This method uses the geometric rules of the exploration region and sequential Monte Carlo programming to select the final target point. This method effectively increases the moving distance of the robot in a single exploration, solves the local problems of the forward simulation method, and improves the efficiency of autonomous exploration. The comparative experiments of G-FS, forward simulation and boundary scanning methods show that the proposed G-FS algorithm greatly reduces the amount of calculation, significantly reduces the mapping time and has high efficiency in the same environment.