论文部分内容阅读
随着科技发展,激光雷达在无人驾驶车中的应用成为社会热门的话题。其中,栅格地图也成为动态障碍物检测的手段之一。原始贝叶斯推理方法在栅格概率趋于极值时,若栅格状态发生改变则体现出的很强的滞后性,因此提出一种利用模糊逻辑矫正权值变量对贝叶斯后验概率进行限制的算法。应用改进的贝叶斯推理更新栅格状态并利用冲突变量检测动态障碍物。最后,通过膨胀、腐蚀、改进连通区域标记法及一维数据区间密度算法提取障碍物信息及可行驶区域信息。实车实验表明提出方法的有效性。
With the development of science and technology, the application of laser radar in driverless vehicles has become a hot topic in society. Among them, the grid map has also become one of the means of dynamic obstacle detection. The original Bayesian inference method has a strong hysteresis if the grid state changes when the probability of the grid tends to the extreme value. Therefore, a method of using the fuzzy logic to correct the weight variable to the Bayesian posterior probability Limit the algorithm. Apply improved Bayesian inference to update the grid state and use the conflict variables to detect dynamic obstacles. Finally, obstacle information and driving area information are extracted by swelling, corrosion, improved connectivity area labeling and one-dimensional data interval density algorithm. Actual vehicle experiments show the effectiveness of the proposed method.