Implementing GMapping with Two Lidars

Question

So, I am attempting to autonomously navigate a Clearpath Robotics' Husky using GMapping. The Husky is currently equipped with two Sick LMS151 lidars, one centered on both the left and right side. Presently, the laser data from the two Sick LMS151 lidars is being merged together using the ira_laser_merger package available at https://github.com/iralabdisco/ira_laser_tools. Unfortunately, while this package does merge the laser data, it does so in a way that creates false data at different points around the Husky. As an example, when I stand in front of the Husky and move backwards, it shows this. However, it also shows me moving toward the back of the Husky. Is there a way to fix this problem using the packages I am currently using or is there a better method I can implement for using two lidars with GMapping?

---

Originally posted by CMobley7 on ROS Answers with karma: 28 on 2015-06-24

Post score: 0

---

Original comments

Comment by asimay_y on 2016-02-02:
hi, have this solved?

Comment by akshaywifi on 2020-07-20:
i am facing same issue . i already use this methods but did not get proper solution map . is there any other package available . or method

Answer 1

First, I suggest ensuring that the LIDARS are accurately centered on both the robot and in the URDF. While the URDF showed our LIDARS were centered, they were slightly off on the actually robot. Due to how we affixed the LIDARS to the robot, it was easier for use to fine-tune the placement of the LIDARS sensors in the URDF. The LaserScan data from each LIDAR was shown in RVIZ, each in a different color. We performed multiple iterations of placing an object at different distances and slightly modifying each LIDARS position in the URDF until they agreed as to the placement of the object. However, this could easily be done the other way around, slightly adjusting the placement of the LIDAR sensor on the robot until they agree. This was successful in removing most of the false data. So, depending on your application this may be the optimal solution. In addition, I suggest using many of the filters in laser_filter to filter the raw LaserScan data to ensure your are giving quality data to GMapping. Another option would be to use the laser_filter package to limit the range of the Sick LMS151 sensor to 180 degrees and add another sensor for front-facing FOV. This last option was what we choose to implement, due to the fact that the LIDARs would not have more than 180 degree FOV in our final application.

---

Originally posted by CMobley7 with karma: 28 on 2016-02-03

This answer was ACCEPTED on the original site

Post score: 0