【计算机科学】【2016.09】三维点云中的目标检测

本文为德国弗里恩大学（作者：ChristianDamm）的硕士论文，共54页。

随着自动驾驶车辆的不断普及，避障等挑战变得越来越重要。为了实现避障，可靠的障碍物检测是前提条件之一。虽然普通的自动驾驶车辆主要使用相机和雷达传感器来实现这一目的，但是目前激光测距传感器正在逐渐作为可实施的替代方案。由于激光传感器的精度很高，在不同的工业领域中得到了广泛的应用。通常，传感器数据被用作点云的形式。在本篇硕士论文中，提出了一种基于点云的障碍物检测方法。因此，进行可靠障碍检测的几个子任务，如下采样和平面分割等得到了实现。最后，提出了一种基于线性卡尔曼滤波的障碍物跟踪算法，在自动驾驶车辆MadeInGermany（MIG）的多个测试驱动器内进行了实验评估。

With the ongoing spread of autonomousvehicles, challenges like obstacle avoidance get more important. To realizeobstacle avoidance, a reliable obstacle detection is one of the preconditions.While common autonomous vehicles mainly use camera and radar sensors for thispurpose, currently laser range sensors are enforcing as alternatives. Due toits high accuracy, this kind of sensor establishes in different industries. Ingeneral, the sensor data is used as point clouds. Within this master’s thesis,an approach for obstacle detection based on these point clouds is presented.Therefore, several subtasks, e.g. downsampling and plane segmentation, of areliable obstacle detection are carried out. Finally, an algorithm for obstacletracking, based on a linear Kalman filter, is implemented. The received resultsare evaluated within several test drives of the autonomous vehicleMadeInGermany (MIG).

1 引言
1.1 研究动机
1.2 研究目标
1.3 论文结构
1.4 相关工作回顾
2 基础知识
2.1 MIG
2.2 LiDAR
2.3 点云
2.4 K-d树
2.5 ROS
2.6 PCL
3 具体实现
3.1 下采样
3.2 地面分割
3.3 聚类
3.4 跟踪/卡尔曼滤波器
4 实验评估
4.1 场景设置
4.2 实验结果
5 结论

下载英文原文地址：

http://page5.dfpan.com/fs/1lcfj2d21029b160031/

更多精彩文章请关注微信号：