Error initializing Outlier Removal Filter using make_statistical_outlier_filter()

[rdspring1]

outlier_filter = pcl_data.make_statistical_outlier_filter() Error: TypeError: __cinit__() takes exactly 1 positional argument (0 given)

[rdspring1]

There is a mistake in the python-pcl build. The constructor for the filter requires the input point cloud as an argument, but the python interface is not passing the argument.

Here is my fix: Summary: Remove the unnecessary argument and then reinstall python-pcl.

Part 1:

In pcl/pxi/Filters/StatisticalOutlierRemovalFilter_172.pxi

Change

def __cinit__(self, PointCloud pc not None):
         self.me = new pclfil.StatisticalOutlierRemoval_t()
         (<cpp.PCLBase_t*>self.me).setInputCloud (pc.thisptr_shared)

to

def __cinit__(self):
         self.me = new pclfil.StatisticalOutlierRemoval_t()

Repeat for all instance of the constructor in file

Part2:

In pcl/pxi/PointCloud_PointXYZ_172.pxi Change

def make_statistical_outlier_filter(self):
    # fil = StatisticalOutlierRemovalFilter()
    # cdef pclfil.StatisticalOutlierRemoval_t *cfil = <pclfil.StatisticalOutlierRemoval_t *>fil.me
    # cfil.setInputCloud(<cpp.shared_ptr[cpp.PointCloud[cpp.PointXYZ]]> self.thisptr_shared)
    return StatisticalOutlierRemovalFilter(self)

to

def make_statistical_outlier_filter(self):
      fil = StatisticalOutlierRemovalFilter()
      cdef pclfil.StatisticalOutlierRemoval_t *cfil = <pclfil.StatisticalOutlierRemoval_t *>fil.me
      cfil.setInputCloud(<cpp.shared_ptr[cpp.PointCloud[cpp.PointXYZ]]> self.thisptr_shared)
      return fil

Part 3:

1) Delete pcl/_pcl_172.cpp It is auto-generated by cython. The build process skips this step if it is already present

Part 4:

Reinstall python-pcl

[ryan-keenan]

Thanks for debugging this! I've tested and confirmed that this fix works and pushed the change to the RoboND-Perception-Exercises repo.