基于激光雷达的动态障碍物实时检测

动态障碍的存在直接影响到环境地图的构建精度,可靠实时地检测出动态障碍物是未知环境下移动机器人构建环境地图的根本前提。基于2D激光雷达传感器,提出了一种移动机器人在未知环境下实时检测动态障碍物的方法。将激光雷达的观测数据经过滤波映射到世界坐标系,构建相邻采样时刻的三幅栅格地图;判断相邻时刻三幅栅格地图上对应栅格的占用状态,确定环境中的静态障碍物,以静态障碍物为参考,根据当前的栅格地图可以检测出环境中的动态障碍物。基于激光雷达时空关联性分析,采用八邻域滚动窗口的方法处理不确定性因素。在实际移动机器人MORCS-1上进行的实验结果表明,该方法可使移动机器人准确有效地检测出未知环境中的动态障碍物,实时性好,可靠性高。

Real-time Detection of Dynamic Obstacles Based on Laser Radar

Presence of dynamic obstacles directly affects the precision of building environment maps.Real-time and reliable detection of dynamic obstacles in unknown environments with mobile robots is very essential for building environment maps.An approach to detecting dynamic obstacles in unknown environments for mobile robots in real-time is proposed based on 2D laser radar as the external navigation sensor.Sensor measurements with noisy data filtered out are firstly mapped into the world coordinate system to build three grid maps at the three consecutive sampling moments.Static obstacles are then detected by comparing the states of the corresponding grid cells in the three consecutive grid maps.With static obstacles detected as a reference,dynamic obstacles are lastly identified according to the current grid map.Based on the analysis of space-time correlation of laser radar,the uncertainty is handled using the sliding window of the surrounding eight grid cells.Experiments are conducted in a physical mobile robot MORCS-1,and the results show that the mobile robot can detect dynamic obstacles accurately and effectively in unknown environments with good real-time performance and high reliability.