Re: rtabmap mapping mapping publish strange tf

Posted by matlabbe on
URL: http://official-rtab-map-forum.206.s1.nabble.com/rtabmap-mapping-mapping-publish-strange-tf-tp7300p7303.html

What is the twist covariance set in odom_combined ? With repetitive textures, rtabmap is likely to find wrong loop closures. The only way for rtabmap to reject those loop closures is to have a covariance in odometry not over estimated.

Note that with

<!-- If set, TF is used to get odometry instead of the topic -->         
<arg name="odom_frame_id"           default="odom_combined"/>

You should set

<arg name="odom_tf_angular_variance" default="1"/>             <!-- If TF is used to get odometry, this is the default angular variance -->
<arg name="odom_tf_linear_variance"  default="1"/>             <!-- If TF is used to get odometry, this is the default linear variance -->

to match your odometry covariance. With wheel odometry, you could set them as:

<arg name="odom_tf_angular_variance" default="0.005"/>             <!-- If TF is used to get odometry, this is the default angular variance -->
<arg name="odom_tf_linear_variance"  default="0.001"/>             <!-- If TF is used to get odometry, this is the default linear variance -->

Enable graph check rejection (this can detect wrong loop closures):

<param name="RGBD/OptimizeMaxError"           type="string" value="3"/>

cheers,
Mathieu