What effect does the low msg update rate do to the SLAM map ?

asked 2016-04-07 13:41:08 -0500

alice222777 gravatar image

updated 2016-04-08 05:51:19 -0500

hi

i'm building the map with Hector SLAM and my customized lidar. The structure of customized lidar is just like RPLidar, it's always rotating.

image description

▼ I got a good scanning result when i placed the lidar in a single point.

(there's a big window, so the down side of the map looks like that, it's correct)

image description

▼ but the map sometimes failed.

image description

 

 

 

I have no idea why the system is so weak that once if i moved or rotated the heading of lidar too fast with hands, the map failed easily. After checking some of the Q&A here, and i found this, here's some questions about the problem happened on my system :


  • Is it caused by the low msg update rate too ? Or is it caused by some other problems ?

  • If it's not caused by the low msg update rate, what effect does the low msg update rate do to the SLAM map ?

  • what's the minimum angle resolution or the basic needs for Hector SLAM that can make the scan matching work successfully ?

  • it is that the angle should always be 0~ 2pi in a single msg ? is 0~ pi enough for Hector SLAM to make a map ?  


- thanks ! :)  

 



here's some info of my system

  • LaserScan.msg publishing frequency : about 3.1 hz
  • Sampling rate : 2000 (sample / sec) in total
  • rqt :

   image description

  • LaserScan.msg sample :
header:    
 seq: 31   
 stamp: 
  secs: 1459761437
  nsecs: 218499461   
 frame_id: laser_frame 
angle_min: 0.0 
angle_max: 6.25863647461 
angle_increment: 0.00892857182771
time_increment: 0.000553450663574
scan_time: 0.371918857098
range_min: 0.10000000149
range_max: 40.00
ranges: [ ... ] ( about 590~670 data in this vector ) 
intensities: []

                    


   

edit retag flag offensive close merge delete

Comments

One of the core assumptions of hector_slam is that the world does not move much between laser scans. With a 30Hz scanner like the UTM30-LX, this works fairly well. For lower rates you'll need to move the scanner more slowly.

ahendrix gravatar image ahendrix  ( 2016-04-07 13:53:39 -0500 )edit

oh 30 Hz ? thank you :) i'll make some test to see what will happen if the msg publishing frequency is higher.

alice222777 gravatar image alice222777  ( 2016-04-08 05:46:52 -0500 )edit

The publish rate on the UTM30-LX is a direct result of their rotation rate; they publish one message per spin. Unless you change the rotation rate of your scanner, changing the publish rate probably won't have the desired affect.

ahendrix gravatar image ahendrix  ( 2016-04-08 17:51:19 -0500 )edit

ah, okay, i'll make the device rotate faster :) thanks for telling me so much things

alice222777 gravatar image alice222777  ( 2016-04-09 10:00:23 -0500 )edit