导航

Multi-sensor Fusion Localization

Introduction

The goal of multi-sensor localization is to provide a robust method which can achieve high localization accuracy and resilience in challenging scenes, such as urban downtown, highways, and tunnels. It adaptively uses information from complementary sensors such as GNSS, LiDAR and IMU. If you want to know more details, please refer to our paper Guowei Wan, Xiaolong Yang, Renlan Cai, Hao Li, Yao Zhou, Hao Wang, Shiyu Song. “Robust and Precise Vehicle Localization Based on Multi-Sensor Fusion in Diverse City Scenes,” 2018 IEEE International Conference on Robotics and Automation (ICRA), Brisbane, QLD, 2018, pp. 4670-4677. doi: 10.1109/ICRA.2018.8461224. link

Currently this localization method only works for x86_64 platform

Input

  • Point cloud data from LiDAR sensor (ROS topic /apollo/sensor/velodyne64/compensator/PointCloud2)

  • GNSS observation and ephemeris data from GNSS sensor (ROS topic /apollo/sensor/gnss/rtk_obs and /apollo/sensor/gnss/rtk_eph)

  • GNSS best pose from GNSS sensor (ROS topic is /apollo/sensor/gnss/best_pose)

  • Imu data from IMU sensor (ROS topic /apollo/sensor/gnss/imu)

  • Localization map (FLAGS_map_dir + “/” + FLAGS_local_map_name)

  • Parameter config files (velodyne64_novatel_extrinsics_example.yaml, velodyne64_height.yaml, and ant_imu_leverarm.yaml, located in modules/localization/msf/params/)

Output

  • Localization result defined by Protobuf message LocalizationEstimate, which can be found in file localization/proto/localization.proto. (ROS topic /apollo/localization/pose)

  • Localization status defined by Protobuf message LocalizationStatus, which can be found in file localization/proto/localization.proto. (ROS topic /apollo/localization/msf_status)

  • We also provide intermediate results: Gnss localization result, LiDAR localization result. (ROS topic /apollo/localization/msf_gnss, /apollo/localization/msf_lidar)

Modes

MSF Localization Modes

We provide 4 modes for msf localization module. 3-Systems uses GNSS localization result all the time, while 2-Systems only apply GNSS localization result to initilize SINS alignment.

  1. 3-Systems(BestGnss + LiDAR + SINS): gnss_mode(0), gnss_only_init(false).

  2. 3-Systems(Local-Gnss + LiDAR + SINS): gnss_mode(1), gnss_only_init(false).

  3. 2-Systems(BestGnss + LiDAR + SINS):gnss_mode(0), gnss_only_init(true).

  4. 2-Systems(Local-Gnss + LiDAR + SINS): gnss_mode(1), gnss_only_init(true).

GNSS Localization Setting

This module works in two modes. One is directly to use GNSS best pose (/apollo/sensor/gnss/best_pose) from INS system, such as NovAtel system. In the second mode, the initial gnss observations and ephemeris information (/apollo/sensor/gnss/rtk_obs and /apollo/sensor/gnss/rtk_eph) are used to calculate GNSS localization result.

You can set gnss_mode in apollo/localization/conf/localization.conf to decide which mode you want to use. The default mode is gnss best pose.

LiDAR Localization Setting

We provide three modes of LiDAR localization: intensity, altitude, and fusion. You can set the parameter lidar_localization_mode in file localization.conf to choose the mode. Considering computing ability of different platforms, we provide lidar_filter_size, lidar_thread_num and point_cloud_step to adjust the computation cost. Furthermore, we provide three yaw optimization methods in LiDAR localization algorithm. You can set the parameter lidar_yaw_align_mode in file localization.conf to choose the mode.

Generate Localization Map

Localization map is used for LiDAR-based localization, which is a grid-cell representation of the environment. Each cell stores the statistics of laser reflection intensity and altitude. The map is organized as a group of map nodes. For more information, please refer to apollo/modules/localization/msf/local_map.

We provide a script (apollo/scripts/msf_local_map_creator.sh) to generate localization map. Usage: bash msf_local_map_creator.sh [pcd_folder] [pose_file] [zone_id] [map_folder].
pcd_folder: contains all pcd files(as .pcd file) which are used to create map.
pose_file: a file which contains all poses of pcd files. Each line corresponds to a pcd file, and the format of each line is “pcd_number timestamp x y z qx qy qz qw”. Thereinto, “pcd_number” corresponds to the pcd file name.
zone_id: UTM zone id.
map_folder: destination folder for map.

After running this script, you will get a localization map in map_folder, which contains a “config.xml” and “map” folder。 Actually, in this script, we first create lossless map and then convert lossless map to lossy map which losses some not essential message and has smaller size. The lossless map is also avaliable.

Visualization Tool

We provide a simple online visualization tool for debug multi-sensor fusion localization module. The parameter use_visualize in localization.conf is used to enable the online visualization tool.

We also provide an advanced online visualization tool which can show localization map, point cloud, horizontal position and variance circle of LiDAR-localization, GNSS-localization, and fused localization results. This tool is simply launched by ./scripts/localization_online_visualizer.sh.

Check MSF Localization Status

We provide a simple way to check lidar localization, GNSS localization and fusion localization status. There are four states {NOT_VALID, NOT_STABLE, OK, VALID} for localization status. You can simply use rostopic echo /apollo/localization/msf_status to check localization status. If fusion_status is VALID or OK, the output of msf localization is reliable.

上一页

定位

下一页

NDT-based Lidar Localization