hector_mapping map overlap

asked 2013-03-29 02:04:41 -0500

elva
1 ●4 ●5 ●7

updated 2013-03-29 02:08:45 -0500

I am doing hector mapping using a hokuyo URG-04LX ,i hold the hokuyo in hand,moving it slowly ,but the mapping result is not very good,it is not accurate and clear,always overlaps. My launch file is following the mapping_default.launch which is inside the hector mapping.And i use only hokuyo_node and hector_mapping node .the tf seems to be ok,the tf tree is : 1. -/base_link → /laser 2. -/map → /scanmatcher_frame

Is there any way to solve the problem?

edit retag flag offensive close merge delete

add a comment

2

answered 2013-03-29 03:56:23 -0500

Stefan Kohlbrecher

24361 ●173 ●299 ●444 https://www.energy-rob...

The URG-04LX has pretty low range and low scan rate, so fast movements and environments with few features (in range) present difficulties. As you don't appear to use an IMU, it is also important to always keep the sensor's scan plane parallel to the ground plane at all times.

edit flag offensive delete link

Comments

thank you for your reply^^i have an microstrain imu, is it correct that i add hector_imu_attitude_to_tf _node and microstrain_3dmgx2_node into my launch file? In addition, I found a video which perform what i want to do ,http://www.youtube.com/watch?v=quqF5_ZE_fI,he also use 5.6m 10hz 2D lidar.

elva ( 2013-03-29 20:38:17 -0500 )edit

May i know how the tf works?when i added imu and hector_imu_attitude_to_tf ,besides the previous 2 tf there is one more tf : 3. -/base_stabilized → /lase_footprint. but there is warning: no transform from base_footprint to map and No transform from [/base_stabilized] to frame [/map].how to fix it?

elva ( 2013-03-29 22:25:47 -0500 )edit

concerning the youtube video: keep in mind that it shows simulation. If you take a look at the real robot videos you can see that the map looks much worse on a real robot. Furthermore the videos don't show vanilla hector slam, but some modified version, i.e. scan matching with a 3D not 2D map.

jodafo ( 2013-04-01 22:59:50 -0500 )edit

I am doing a project quite similar to that video in real world, i have a hokuyo(URG-04LX),on which there is a mirror for height estimation, i also have a imu(3DM-GX3-25) attached to my hokuyo,since i want to apply themon a quadrotor to do 2.5 mapping using octomap sever.but the 2D map overlaps.

elva ( 2013-04-02 16:04:46 -0500 )edit

add a comment

0

answered 2013-11-21 21:05:44 -0500

Yuichi Chu
148 ●9 ●12 ●17

Hi,elva I have met the same problem.I use the hector_slam,but the map is mess.Have you solved the problem?Can you give me some advice?Thank you very much.

edit flag offensive delete link

add a comment

0

answered 2013-04-01 23:13:40 -0500

jodafo

365 ●3 ●7 ●12

updated 2013-04-01 23:14:51 -0500

Hi, you said "the tf tree is : 1. -/base_link → /laser 2. -/map → /scanmatcher_frame"

I'm not sure of the notation, but that doesn't seem to be correct? /map should at least link to /base_link, and /base_link to /laser, where you have to provide the transforms for /base_link to /laser.

However, I'd recommend following the tutorial (http://www.ros.org/wiki/hector_slam/Tutorials/SettingUpForYourRobot), i.e. /base_footprint -> /base_stabilized -> /base_link -> /laser.

Since you don't have odometry, both base_frame and odom_frame parameters should be set to /base_footprint. Dunno if hector_slam also works with base_frame and odom_frame set to /base_link.

edit flag offensive delete link

Comments

Really thank you for replying^^ The frames problem i have solved,however, the laser scan still can not match well, therefore the map overlaps sometimes, i am now trying to use the imu data :yaw angle to help my hokuyo works better ,any help?thank you:P

elva ( 2013-04-02 15:51:01 -0500 )edit

add a comment

hector_mapping map overlap

3 Answers

Comments

Comments

Question Tools

Stats

Related questions

hector_mapping map overlap edit

3 Answers

Comments

Comments

Question Tools

Stats

Related questions

hector_mapping map overlap