I have a freshly new Rplidar 2 that I'd like to use to build some maps with my basic iRobot2 create. I tried using `gmapping`, but apparently there are inconsistencies between the `rplidar` node and `gmapping` itself (that can be solved for Rplidar 1 following [this link](http://geduino.blogspot.co.uk/2015/04/gmapping-and-rplidar.html), but at a first sight, it doesn't work with Rplidar 2 - I'll open a separated question for this). Anyway, I decided to use Hector Slam, as "adverstised" on the Rplidar rosWiki page (actually the `hector_mapping` node of Hector Slam). I managed to build some basic maps, mainly without using the odometry, but what I noticed is:
- The `hector_mapping` node often crashes with the recurrent message `SerachDir angle change too large` (I searched around and there are some solutions, like [this](http://answers.ros.org/question/218627/how-to-fix-problem-searchdir-angle-change-too-large/))
- The produced maps are really coarse grained (almost at the level of the ones I was obtaining with a kinect). In many videos and threads around I saw many maps done with the Rplidar that are much more defined than the one I'm getting.
Now, given that, my general question is: does anybody have accumulated enough experience to propose some really good configuration to use with hector_slam and Rplidar 2 on a wheeled robot, where for configuration I mean any possible configurable parameter (including publishing frequency of involved static transformation publishers)?
↧
