You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Norlab_icp_mapper currently throws an error when the ICP doesn't converge to a valid solution. Subsequently, this package catches the error, saves the current map and exits. Although legitimate during offline mapping sessions, this approach is not very practical in real-life mapping applications. It is therefore currently overcome by setting the limits for the convergence error to very high values. Here's an example from Warthog's config file:
Instead, this package could catch the error, log it and use the odometry estimate without the ICP correction. Furthermore, users could use a parameter defining the strictness of the convergence error handling.
The text was updated successfully, but these errors were encountered:
The reason why the warthog config is set to these high values is because of some other bug we had when loading the initial map in the teach and repeat node. I understand what you mean, but I think that it is fine if this exception causes the mapping to crash. It means that there are problems that need to be addressed in the ICP config.
Norlab_icp_mapper currently throws an error when the ICP doesn't converge to a valid solution. Subsequently, this package catches the error, saves the current map and exits. Although legitimate during offline mapping sessions, this approach is not very practical in real-life mapping applications. It is therefore currently overcome by setting the limits for the convergence error to very high values. Here's an example from Warthog's config file:
Instead, this package could catch the error, log it and use the odometry estimate without the ICP correction. Furthermore, users could use a parameter defining the strictness of the convergence error handling.
The text was updated successfully, but these errors were encountered: